Skycap articles and field notes

How long checked bags take, and why nobody measures it

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

No US airport or airline publishes how long bags take. The closest thing is a promise: Delta and Alaska both say 20 minutes, and they pay miles when they miss. My working assumption is that your own bag reaches the belt about 20 minutes after the plane parks at a typical airport, 23 at a large hub and 30 off a widebody. Those are judgement, and I’ll show you how I got there.

Why the number doesn’t exist

Bags come off the plane in a process run by airlines and their handlers, and the scan that says a bag reached the belt stays with them. DOT publishes how many bags get mishandled. It doesn’t publish how fast the good ones arrive. Even IATA says the delivery metric isn’t universally measured.

I searched public records, airport reports and government documents through 2026 and found no US first-bag or last-bag distribution. I’d be glad to be corrected.

What has been promised

Who	Promise	If they miss
Delta	20 minutes on domestic flights	2,500 miles, reported claim window of 2 hours since 2026
Alaska	20 minutes after the plane parks	$25 discount code or 2,500 miles, request within 2 hours

A promise tells you what an airline believes it can usually hit. It tells you nothing about how often it does. Reports differ on whether Delta’s clock starts at gate arrival or at the door, and I couldn’t read Delta’s own terms page, so check it before you claim.

Airports’ advice to people picking you up gives a rough cost for the bag. Orlando says 20 minutes without bags, 40 with. Los Angeles says 15 and 30. So they put it at 15 to 20 minutes more.

What other countries measure

Italian airports publish bag times in their service charters, on a clock starting when the plane blocks in. Gatwick publishes monthly. These are the closest figures to what you’d want for the US.

Airport	Measure	Result
Cagliari, 2023	First bag, 90th percentile	19.3 min
Turin, 2024	First bag, 90th percentile	20.4 min
Pisa, 2025	First bag, 90th percentile	21.2 min
Milan Malpensa T1, 2023	First bag, 90th percentile	25.5 min
Milan Malpensa T1, 2023	Last bag, 90th percentile	39.8 min
Gatwick, Aug 2025, smaller aircraft	Flights with last bag inside 40 min	95.67%

In the US, the one share I could find is second-hand: an American Airlines executive said in 2021 that 63% of bags at Miami came within 20 minutes, with no clock stated. Treat it as a hint.

The airports above are mostly smaller than a US hub, and US hubs have long tug runs. An airport planning study, ACRP Report 23, also notes that hub operators often appear to favour connecting bags over local ones.

The assumptions I use

Case	First bag, typical and 90th percentile	Your bag, typical and 90th percentile
Narrowbody, typical airport	15, 26	20, 31.5
Narrowbody, large hub	17, 30	23, 36
Widebody	19, 32	29.5, 47.5

Minutes run from the plane parking. Your own bag lands somewhere between first and last, which is what most pages leave out: “first bag” tells you when the belt starts, and you may be the last person there. The own-bag column is a simulation with a chosen correlation between first and last bag. It’s a judgement built on a judgement, and I’d replace it the day someone publishes real numbers.

Your bag is racing you

The cost of checking a bag depends on your walk. If you reach the belt after your bag does, you’ve lost nothing. At Burbank, a small terminal, a bag adds about 13 minutes to the typical trip to the curb, because you get to the belt quickly and wait. At Atlanta’s Concourse E, in this worked example, it adds about 2, because you’re on the train while the bag gets there.

So skip the “add 15” rule. With a long walk, a checked bag costs almost nothing in minutes. With a short one, it’s most of your wait.

If it’s late

Note the time the plane parked. On Delta or Alaska, past 20 minutes, claim within 2 hours, and check the current terms first. Oversize items are generally excluded. If the belt stops and your bag isn’t on it, go to the airline’s baggage office before you leave the airport.

If someone’s waiting, have them hold in the cell phone lot until the bag is in your hand. To see when the airline thinks you parked, compare what the tracker and the airline each report. The bag term also feeds the airport figures at Landing to the curb, by airport.

Tell Skycap you checked a bag and your curb time changes by what it costs at your airport and your gate. If you haven’t said, it shows both. How at the curb works

How long from landing to the curb

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

Plan on about 20 to 35 minutes from touchdown to the curb at a big US airport with a carry-on, and about 30 to 45 with a checked bag. Those are typical times, worked out on Nov 7 2026. The slow end runs further: at O’Hare with a checked bag it reaches about 68 minutes.

Airports’ own rules of thumb are short of that. Los Angeles says about 15 minutes with no checked bags, 30 with them and 45 from abroad. Orlando tells people meeting a flight to allow at least 20 minutes, or 40 with bags. My typical figure for LAX is 27 with a carry-on. Fifteen is someone near the front of the plane with a near gate, and most of the plane isn’t.

Six stages, and they aren’t equal

Stage	Size	Basis
Taxi to the gate	7 min national median, 16 at the 90th percentile. Midway 4, O’Hare 18	Measured
Door	About 1 min	Judgement
Getting down the aisle	About 3 min from the front, 10 from row 30	Modelled
Walk and train	2 to 4 min at Burbank, 3 to 30 at Dulles, 2 to 30 at LAX	Computed from maps
Your checked bag	About 20 min after parking, 23 at a large hub	Judgement
Passport control	US passport median 5 at Atlanta, 18 at Orlando	Measured hourly averages

Taxi-in gets the attention because it’s the stage you can watch. It’s rarely the one that decides your evening. Your row and your bag move the answer more, and you know both before you land.

Typical times, by airport

Times are minutes from touchdown to the arrivals curb on a domestic flight, for an unknown row and gate on an A320-sized plane about 82% full. The range is meant to hold about 85 trips in 100 on a normal day. That’s a design target, and nobody has timed it.

Airport	Carry-on	Checked bag
Burbank	13 (6 to 22)	26 (15 to 37)
Midway	19 (11 to 28)	28 (19 to 38)
Boston	21 (13 to 32)	31 (21 to 43)
LAX	27 (13 to 43)	37 (25 to 52)
JFK	28 (16 to 42)	37 (25 to 51)
Atlanta	32 (20 to 46)	38 (27 to 50)
Denver	33 (20 to 48)	39 (28 to 53)
O’Hare	34 (19 to 56)	45 (29 to 68)

From abroad, add passport control and a widebody’s slower bags. Landing at JFK with a checked bag comes to about 46 minutes on a US passport (32 to 62) and about 57 on another passport (33 to 82). Those use the airport’s domestic taxi figures, because nobody publishes international ones.

One plane, two concourses

Take a Delta 737-800 landing in Atlanta at 2 PM in June, and you in row 25. Taxi is about 7 minutes and getting off is about 9, door included, so you’re on the jet bridge about 16 minutes after touchdown wherever you park.

At a Concourse T gate, you’re near the terminal. The curb comes at about 24 minutes (18 to 31). At Concourse E, you ride the train and walk, and the curb comes at about 38 (29 to 48).

Now check a bag. At T that takes you to about 34 minutes, 10 more. At E it takes you to about 40, only 2 more. You arrive at the belt after your bag does, so the walk has already spent the wait. This is why a flat “add 15 for bags” is wrong in both directions, and why an airport’s single number can’t be right for every gate.

How the numbers were made

Taxi-in comes from a year of public flight records, Sep 2025 to Aug 2026, as medians and 90th percentiles by airport, carrier and local hour. Deplaning is a model fit to 30 timed flights. Walks are computed along mapped paths at 1.34 m/s, with 1.0 m/s for the slow end. Passport figures are hourly averages from public wait-time reports, 12 months to Nov 4 2026.

The stages combine by adding variances, because adding every 90th percentile assumes everything goes badly at once. Bags are a race: you leave claim at the later of reaching the belt and your bag reaching it.

Nobody has timed this

Nothing here is a timed gate-to-curb trip. Published airport walking times run about 1.4 times my median walker, so I raised the slow end to match. That means the check isn’t fully independent.

Bag times are judgement, because the US has no public measurement. Stages are treated as independent, and bad days stretch several together, so real tails run longer. A flight landing 30 minutes early may also wait for its gate, which these figures don’t include. Airport tables describe a normal day, and closures aren’t in them.

If your landing time itself looks wrong, the tracker and the airline may be timing different events. The full method, with other airports, is at Landing to the curb, by airport.

Skycap runs this chain for your flight, with your row, your bag and your gate once it’s known. The range tightens as each stage finishes, so you can see when you should be at the curb. How at the curb works

How long it takes to get off the plane, by row

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

On a 737 or A320 at a normal load, expect about 3 minutes from the front rows and about 10 from row 30, counted from the door opening. Add about a minute between parking and the door. That minute is my judgement, with a range of 0 to 2.

The usual advice is 30 seconds a row, and it comes from somewhere real: one flyer who timed himself. It’s a good rule in the middle of the cabin. It’s short at the front and long at the back, because the first few minutes are mostly people standing up and getting bags down, and no row moves during that part.

The table

These figures are for Delta’s 737-800, 160 seats, 82% full, in a middle seat, with a typical chance of an overhead bag. Minutes are from door open to stepping off.

Row	Seats ahead	Typical	10th to 90th percentile
4	12	2.9	1.6 to 4.3
10	16	3.1	1.8 to 4.5
15	46	4.8	3.4 to 6.3
20	76	6.5	4.9 to 8.1
25	106	8.2	6.4 to 9.9
30	136	9.9	7.9 to 11.7

Look at rows 4 and 10. They’re almost the same, because Delta’s numbering skips from 4 to 10. Row 10 is the fifth row on that plane. Your row number is a label, and what counts is how many seats sit between you and the door. The timed flights say the same: physical row predicted better than ticketed row, with an average error of 0.94 minutes against 1.23.

What moves it, and by how much

A full flight costs more the further back you sit. At row 30, a plane 95% full takes about 11.1 minutes and one 65% full takes about 8.3. At row 10 the difference is under half a minute.

An aisle seat saves 0.4 to 1.1 minutes against a middle seat, and more from the back. A window costs a few seconds. Paying for an aisle to get off sooner is a poor trade. If you want those minutes, the seats ahead are what buy them: moving from row 25 to row 15 saves about 3.4.

The whole span from row 10 to row 30 is about 7 minutes. A checked bag usually costs more than that, and at O’Hare in the evening, taxi-in can too.

One traveller’s 30 flights

Nearly everything public about row timing traces to one traveller, who timed 30 of his own flights from Aug 2016 to Feb 2018, mostly on A320s and 737s. He reports about 29 seconds per physical row. Those flights fit a line better than a flat rate: about 1.7 minutes plus 22 seconds a row.

I built a simulation with seats, bags and aisles, tuned it to a 15 to 17 passengers a minute rate reported in a 2014 airline study, and scored it on each of his flights with that flight held out. Then I compared simple formulas the same way.

Method	Average error on 30 flights
Compact formula used here	0.91 min
Full simulation	0.92 min
Straight line in physical row	0.94 min
30 seconds a physical row	1.13 min

The simulation didn’t beat the straight line. That was the disappointing part of the project, and it’s why the answer here is a short formula. Flights scatter by about 1.1 minutes around any line, so nobody gets much closer than this from a seat number. The 10th to 90th percentile bands held 83% of those flights, close to the 80% they should.

What the formula adds beyond 30 seconds a row is load, layout and seat position, using public load factors from Aug 2025 to Jul 2026. Those extensions rest on the simulation. His 30 flights can’t check them, since he assumed full flights and usually sat in the middle.

Where to distrust it

Widebody and regional numbers are unvalidated. A two-aisle plane with a mid-cabin door behaves differently, and nobody has published the timing. I don’t have row maps for American or United, so their figures use a wider range.

One person with one stopwatch from 2016 to 2018 is thin. Bin habits may have changed since. His clock probably started a little before the door truly opened, which is part of why the door allowance above is uncertain.

Don’t use these numbers for anyone with wheelchair assistance. Assisted passengers usually leave last, so the row doesn’t apply, and the wait for an attendant varies too much to estimate. Regional jets with gate-checked bags add a wait in the jet bridge that nobody has measured.

Getting off is one link in the chain. The rest is in how long from landing to the curb. For the numbers by aircraft family, see how long to get off the plane, by row. If you’re choosing a seat for a smooth ride instead, the smoothest seat is a different trade-off.

Tell Skycap your row, or scan your boarding pass, and it counts the seats ahead of you on your aircraft. Your curb time moves with it. How at the curb works

How long passport control takes, by terminal

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

At JFK, a typical hour takes 9 to 14 minutes on a US passport and 13 to 32 on any other, depending on your terminal. Public wait-time reports for the 12 months to Nov 4 2026 show a bigger gap between JFK’s terminals than between most airports. Terminal 1 is the slow one, and Terminal 5 the quick one.

Those reports come a day at a time. Here they are as a year.

The terminals that differ most

Figures are minutes: the median hour, then the 90th percentile hour. US passports and other passports are separate lines at the booth, so they have separate numbers.

Airport and terminal	US passport	Other passport
JFK Terminal 1	14, 27	32, 69
JFK Terminal 4	10, 16	22, 54
JFK Terminal 5	9, 17	13, 31
JFK Terminal 7	12, 23	22, 41
JFK Terminal 8	10, 16	26, 57
Orlando Terminal C	25, 38	37, 61
Orlando Airside 4	5, 9	14, 32
Miami North	12, 18	21, 43
Miami South	7, 12	16, 39
Miami Central	4, 10	16, 37
LAX Tom Bradley	12, 18	23, 46
LAX Satellite 7	10, 14	15, 32
Newark Terminal B	12, 24	17, 40
Newark Terminal C	10, 17	15, 29
Dulles International A	8, 22	14, 38
Atlanta Concourse E	4, 9	7, 16
Atlanta Concourse F	5, 11	9, 23
Seattle South Satellite	15, 25	25, 54

Orlando has two halls, and a US passport takes five times as long at Terminal C. It’s the slowest big gateway on the median and Atlanta the fastest, but the Orlando figure is mostly one building.

Other gateways with one hall: Dallas/Fort Worth is 9 and 18 at the median, O’Hare 11 and 17, Boston 13 and 19. Nashville is 19 for a US passport, slower than any JFK terminal.

The hour you land changes it more

For six airports I pulled the 90 days to Nov 4 2026 and split them by the hour the flight arrived. At JFK Terminal 1, the median wait for other passports was 59 minutes for flights arriving at 11 AM and 19 at 7 PM. That’s a 40-minute gap in one hall.

Peaks don’t line up from one terminal to the next, because they follow the airline banks each building handles.

Hall	Worst arrival hour, other passports	Median that hour
JFK Terminal 1	11 AM	59
JFK Terminal 8	5 AM	42
JFK Terminal 4	5 AM	33
Orlando Terminal C	6 PM	50
LAX Tom Bradley	3 PM	36
Miami North	10 and 11 AM	31

In those same 90 days, about 1 passenger in 7 at JFK Terminal 1 and about 1 in 5 at Orlando Terminal C waited more than an hour. At JFK Terminal 5, it was under 1 in 100. Those shares come from published counts of passengers by wait band. The 90-day median at JFK Terminal 1 also ran above the year’s: 36 against 32.

Which line you can change

Global Entry and Mobile Passport Control are the two levers. Mobile Passport Control is CBP’s app for US citizens, permanent residents and some visitors, and you submit your details before you reach the hall. Availability varies by airport, so check CBP’s list.

Global Entry users were 8% to 21% of passengers at most of these halls over the last 90 days. CBP suspended the program from late February to March 11 2026 during a partial shutdown, and the US passport average across airports rose only from 12.2 minutes to 13.5. That suggests a modest saving in a typical hour, 5 to 10 minutes at most by my estimate. It’s worth more at Orlando Terminal C and in the bad hours.

The best way to skip the wait is to clear it before you board. CBP lists preclearance at 16 locations, including Dublin, Shannon, Abu Dhabi and the big Canadian airports. A flight from one arrives as a domestic flight, with no booth.

What these numbers are

The reports count passengers on flights arriving in each hour, by how long they took to clear passport control. They leave out the walk to the hall, the bag belt and customs after it. They don’t say exactly where the clock starts, so if you add a walk, you may count part of it twice.

These are hourly averages. A 90th percentile here is a bad hour, not a bad passenger, and people inside that hour waited longer than the average says. Each hour counts once, so a quiet hour with one flight weighs the same as a packed one.

The data is a few days old, and it’s history, so it can’t see today’s staffing. Some halls see few flights, so their figures rest on thinner hours.

The stage sits inside a longer chain: see how long from landing to the curb. If someone’s meeting you, send them when to leave for the airport. The full tables are at Passport control waits, by terminal. If your arrival gate looks wrong on a tracker, here’s why that happens.

On a flight into the US, Skycap adds the typical wait for your terminal and your passport line to your curb time. It leaves that out when you cleared the US border before you left. How at the curb works

Taxi-in: why O'Hare takes 18 minutes and Midway takes 4

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

The median US flight takes 7 minutes from touchdown to the gate, and 1 flight in 10 takes more than 16. O’Hare’s median is 18 and its 90th percentile is 36. Midway, 20 miles away, is 4 and 7. Those figures come from 12 months of public flight records: domestic flights by US carriers, Sep 2025 to Aug 2026.

This stretch is called taxi-in: wheels down to arrival at the gate. Trackers show you “landed”, airlines show you the gate time, and nobody shows the gap, so passengers tend to learn it from the aisle.

Airport	Median	90th percentile
O’Hare	18	36
Dallas/Fort Worth	13	21
Charlotte	11	21
Seattle	11	20
LAX	9	19
JFK	9	18
Orlando	9	18
Houston Bush	9	17
Atlanta	8	16
Denver	8	17
Boston	7	14
San Francisco	6	16
Phoenix	6	13
Minneapolis	5	11
Reagan National	5	15
San Diego	4	13
Baltimore	4	9
Midway	4	7

Size is only part of it

O’Hare is big. Chicago’s aviation department lists more than 7,200 acres, 8 active runways and 199 gates. Midway is about 775 acres with 43 gates.

But look at O’Hare by the hour. These are the medians for the local hour of landing.

Hour of landing at O’Hare	Median	90th percentile
4 AM	8	14
7 AM	19	33
2 PM	18	32
7 PM	23	52
8 PM	23	47
11 PM	10	21

At 4 AM, O’Hare takes 8 minutes, which is about the national norm for a big airport. The same airfield takes 23 at 7 PM. The acres haven’t changed. My read is that distance costs about 8 to 10 minutes there, and the remainder is waiting: for a gate, for a crossing, for the plane ahead. Night flights are few and may use the nearest runway, so treat that split as an inference and not a measurement.

Midway barely moves. Its median stays between 4 and 5 minutes all day. Its busiest hour is quicker than O’Hare’s quietest.

The 90th percentile tells you more than the median here. At 7 PM, 1 flight in 10 spends more than 52 minutes getting to the gate. Across the year, 41.2% of O’Hare’s flights took 20 minutes or more.

The gate is part of the wait

Landing early doesn’t always get you off early. About 13.5% of US flights land 30 or more minutes ahead of schedule, and at that point the gate may still have another plane on it. Across carriers at O’Hare, the chance of running at least 10 minutes over the norm goes from 15.3% for flights landing 0 to 15 minutes early to 22.5% for those landing 30 to 45 minutes early.

NASA’s work on taxi prediction found the ramp, the last stretch by the gates, was the hard part to predict. That matches what you feel in the seat when the plane stops within sight of the terminal.

Carriers differ a little. At O’Hare, Southwest’s median is 12 against United’s 18, and Delta’s is 17. At Midway, Southwest is 4 and Delta 6. The records don’t show the terminal or the runway, so I can’t separate the carrier from where it parks. I’d guess that’s most of it.

O’Hare is also getting slower. Its July median rose from 13 minutes in 2023 to 19 in 2026. I don’t have a cause I can source.

What to do with this

If you land at O’Hare between 7 and 9 PM, add about 15 minutes to whatever you’d assume elsewhere, and more if someone’s waiting. If you have a connection or a car booked, count from the gate time and not from “landed”.

A delay of 15 minutes on the ground isn’t a sign of trouble. A captain’s announcement about waiting for a gate is routine at these airports.

Method and limits

Each flight’s taxi-in comes from public flight records, 12 months to Aug 2026, without cancelled or diverted flights. Percentiles are nearest rank in whole minutes, by local hour of wheels-on. Hours with under 30 flights are dropped.

The records cover US carriers’ domestic flights. They leave out foreign airlines and international arrivals. Regional flights appear under the operating carrier, so a Delta Connection flight flown by SkyWest is counted as SkyWest, and a few regional carriers don’t report at all. New months arrive about 7 weeks late.

Taxi-in is one stage of many: see how long from landing to the curb and how long to get off the plane. The full tables are at Taxi-in by airport, hour and carrier. Why a tracker’s “landed” and an airline’s “arrived” differ is explained in why the tracker and the airline disagree.

Skycap estimates your taxi from your airport, your airline and your hour of landing. It swaps in the real gate time when the plane parks, and your curb time tightens with it. How at the curb works

When to leave to pick someone up at the airport

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

Time your drive to reach the cell phone lot about when the plane lands, then wait there until they text from the door. Most domestic passengers reach the curb 20 to 45 minutes after touchdown, so you’ll wait a while. The lot is the one place at the airport where being early does no harm.

The advice on most pages is to arrive 30 minutes after the scheduled time. It fails twice. The schedule is a poor clock, and 30 minutes is a poor allowance.

Watch the landing estimate, not the schedule

A scheduled arrival is a gate time with padding in it. In a year of public flight records, scheduled arrival missed actual gate arrival by about 26 minutes on average. An estimate made from the real take-off time missed by about 7. So don’t commit to the drive until the plane has left. If they’re on a two-hour flight and you’re 40 minutes out, you have plenty of time to look.

Typical times to the curb, by airport

Minutes after touchdown to the arrivals curb, with an unknown row and gate. The first number is typical and the second is the slow end, meant to cover about 85 trips in 100 on a normal day. Nobody has timed these. The international columns assume a checked bag.

Airport	Carry-on	Checked bag	From abroad, US passport	From abroad, other passport
Atlanta	32, 46	38, 50	43, 59	44, 60
Boston	21, 32	31, 43	45, 60	50, 66
Dallas/Fort Worth	28, 41	40, 54	49, 65	55, 73
Denver	33, 48	39, 53	44, 60	46, 63
JFK	28, 42	37, 51	46, 62	57, 82
LAX	27, 43	37, 52	46, 63	53, 74
Miami	24, 37	35, 50	44, 60	51, 71
Midway	19, 28	28, 38	40, 54	42, 55
O’Hare	34, 56	45, 68	55, 79	60, 85
Orlando	29, 42	35, 48	51, 69	60, 84
Seattle	27, 41	39, 53	50, 66	60, 84

Ask two questions before you set off: did you check a bag, and which passport are you using? Those two answers move the time more than anything else you can find out. On a flight from abroad, ask too whether they’re coming through a terminal with a long line. A row near the back adds a few minutes, as the row table shows.

A worked evening

Say United 1961 from O’Hare should land in Denver about 6:48 PM, your passenger has a checked bag, and you live 35 minutes from the airport.

Denver’s typical checked-bag time is 39 minutes, and the range runs 28 to 53. So they should be outside between about 7:15 and 7:40. Leave home about 6:15, once you’ve seen the plane is on time, and you’re in the lot as it lands. Then you wait for the text, which might come in half an hour. From the lot to the door takes a few minutes, and Denver’s terminal has East and West sides, so ask which one before you pull out.

The text that matters is “I’ve got my bag and I’m at the door”. “Landed” isn’t useful. Neither is “at the gate”. Those two messages are 30 minutes of circling if you act on them.

The curb isn’t a waiting room

Every airport I read says the same thing in different words: keep moving, stay with the car.

Airport	Cell phone lot rule
Boston	30 minutes at most, no unattended cars
Dallas/Fort Worth	Two lots, two hours at most, driver stays with the car
LAX	Open 24 hours, up to two hours
Orlando	Lots available, and cars at the arrivals curb must be actively loading
Phoenix	Meant for 30 minutes or less, stay with your vehicle

Boston and Phoenix are the ones to plan around. Arrive at landing with a 30-minute limit and you’ll hit it before most passengers are out, so at those airports aim for about 15 minutes after touchdown. Reagan National’s site links its cell phone lot as closed, so check before counting on one there. Rules change, and these were read on Nov 7 2026.

Limits

These are estimates from public data and map walks, for a normal day. Bag times are judgement. Passport figures are hourly averages from public wait-time reports for the 12 months to Nov 4 2026, and the international columns use domestic taxi figures, since nobody publishes others. A delay upstream, an occupied gate or a long claim line will all push the time out.

The fuller reasoning is in how long from landing to the curb. If they’re taking a ride instead, here’s where rideshare pickup is. The numbers come from Landing to the curb, by airport.

Ask them to send you a Skycap link before they board. It opens with no account and shows when they should land and when they should be at the curb, using their bag and their row. How sharing a flight works

Where rideshare pickup actually is at the big airports

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

At most big US airports, the ride app pickup isn’t outside baggage claim. Of the 16 airports below, 3 have it on the same level at every terminal. The rest send you upstairs, across a road, into a garage, or at LAX and part of JFK, onto a bus or a train. Your app shows the exact zone. This page tells you what kind of walk to expect before you get there.

I read these from each airport’s own pages on Nov 6 and 7 2026, and nothing here comes from a ride company. Pickup zones move, so go by the signs and your app if they differ from this.

Same level as baggage claim

Airport	Where
Orlando	Terminals A and B: arrivals curb, Level 2. Terminal C: arrivals curb, Level 6
Miami	Arrival level, outside baggage claim
Dulles	Ground level outside baggage claim, on the third curb

Add about 0 to 2 minutes. Orlando also allows the departures curb, Level 3, from 9 PM to 2 AM.

Across the road

Airport	Where
Denver	Level 5, Island 5, East or West side. Book after you have your bags
Phoenix	Terminal 3: south outer curb. Terminal 4: Level 1 outer curbs, by door 1, 2, 6 or 8

Add about 1 to 3 minutes. Denver’s page says to wait until you’ve collected your luggage or reached Level 5 before you request, and I’d follow that everywhere.

Upstairs, on departures

Airport	Where
Dallas/Fort Worth	Upper level curb at each terminal
Charlotte	Departures level, inner lanes, zones 1 to 3
O’Hare	Terminals 2 and 3: upper level, at poles 2A to 2G and 3G to 3K

Add about 1 to 4 minutes. People look for the ride beside the taxis, and it isn’t there. O’Hare’s page names poles only at Terminals 2 and 3, so from Terminal 1 or 5, let your app guide you.

In a garage

Airport	Where
Seattle	Garage, 3rd floor, stalls 1 to 34. Cross a skybridge, then go down one level
Las Vegas	Terminal 1: garage Level 2. Terminal 3: garage Valet Level
Boston	Terminals A, C and E: Central Parking Garage. Terminal B: its garage, Level 2
Minneapolis-St. Paul	Terminal 1: Level 1 of the Green/Gold ramps. Terminal 2: Purple ramp, second level

Add about 3 to 7 minutes, with a checked bag on a lift or two. Seattle says riders should wait for the car to park in a marked stall and shouldn’t accept rides from strangers offering help.

Depends which terminal

Airport	Terminal A	Terminal B	Terminal C
Newark	Across the street	Terminal frontage	Garage, floor 3
LaGuardia	Across the street	Garage, Level 2	Arrivals inner lane

Add 0 to 7 minutes depending on the terminal. Both airports say to check your app.

A bus or a train away

Airport	Where
LAX	LAX-it lot, east of Terminal 1, by shuttle
JFK Terminals 5 and 7	Howard Beach AirTrain station, temporarily, because of construction

Add about 10 to 22 minutes at LAX and 20 to 35 from JFK’s Terminals 5 and 7. If you’re at those two JFK terminals, compare a taxi. JFK’s page says other terminals generally load in front of the building, though that varies by hour.

LAX’s automated people mover has been reported as opening in January 2027, after several delays. I found no statement from the airport on what happens to LAX-it after that, so this entry may change.

The extra minutes on this page are my estimates by class, from walks and waits and not from timed trips. They run from the arrivals curb.

What to do at the airport

Book when you can see where you’re going, not when you land. A driver sent early circles, and a ride you’ve booked from the aisle gets cancelled. Your bag and your walk will take longer than the driver will.

Atlanta and San Francisco aren’t listed because their sites wouldn’t load for me to read, and I don’t list what I can’t read. At an airport I’ve left out, go by your app and the signs.

If someone’s collecting you instead, here’s when they should leave. If your phone says the plane has landed before the crew does, that’s normal.

Tell Skycap how you’re leaving and it names the place to walk to at your airport and adds that walk to your curb time. It can also open your ride app with your destination filled in. How at the curb works

Does the first flight of the day leave on time? 88% do, against 74% later

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

Yes, the old advice holds up. Over the 12 months to August 2026, an aircraft’s first flight of the day left under 15 minutes late 88.1% of the time, against 73.5% for every later flight. Before 9 AM it was 90.9%. Between 5 PM and 8:59 PM it was 66.5%.

If you have a choice and a reason to be somewhere, book the morning flight and set the alarm.

By the hour

This is every US domestic flight in public records for those 12 months, about 7 million of them, by scheduled hour.

Scheduled departure	Left under 15 min late	Left 60+ min late	Cancelled
5 AM	92.6%	3.2%	1.6%
6 AM	92.0%	3.4%	1.9%
7 AM	89.9%	3.8%	1.6%
8 AM	87.6%	4.4%	1.5%
10 AM	82.0%	5.8%	1.6%
Noon	77.2%	7.5%	1.6%
2 PM	72.6%	9.5%	1.7%
4 PM	69.3%	11.2%	2.0%
6 PM	66.4%	13.0%	2.4%
8 PM	64.4%	13.7%	2.5%
10 PM	68.9%	11.2%	2.8%

The slide runs from dawn to about 8 PM with hardly a pause. Each hour is a bit worse than the one before it. The tail matters more than the average: a 6 PM flight is about four times as likely as a 6 AM one to leave an hour or more late.

It isn’t about the clock, though. Of flights before 9 AM, 83% are the plane’s first of the day, and the small group that aren’t do worse: 7 AM flights by a plane that has already flown left under 15 minutes late 85.2% of the time, against 90.7% for first flights.

Why it works

The plane is already there. By my count, the delay reasons on first flights look nothing like the ones later on.

Reported cause	First flight of the day	Later flights
Airline (crew, maintenance and so on)	55%	28%
Late aircraft	16%	45%
National aviation system	22%	21%
Extreme weather	7%	6%

These are shares of delay minutes on flights that arrived 15 or more minutes late. The late aircraft share collapses and the airline share jumps. First flights take away the commonest failure, and what’s left is a crew that isn’t there or a fault found on the walk-around. Those still happen. About 1 in 21 first flights left an hour or more late. Later flights: about 1 in 11.

Fewer first flights get cancelled too: 1.31% against 1.75%.

Where it matters most

The advantage is small at some airports and enormous at others. This compares first flights before 9 AM with all flights leaving 5 to 8:59 PM, at the 30 busiest airports.

Airport	First flights, on time	Evening flights, on time	Gap
Dallas Love	93.4%	49.3%	44 pts
Chicago Midway	92.0%	56.5%	36 pts
Nashville	90.1%	56.4%	34 pts
Orlando	92.2%	58.8%	33 pts
Boston	90.9%	58.4%	33 pts
Denver	92.9%	64.7%	28 pts
O’Hare	88.9%	63.0%	26 pts
Dallas-Fort Worth	85.6%	61.8%	24 pts
Atlanta	89.4%	69.3%	20 pts
Seattle	89.9%	74.5%	15 pts
Los Angeles	89.7%	74.6%	15 pts
Salt Lake City	92.6%	78.7%	14 pts

Love Field’s evening number is the one that surprised me. Fewer than half of its flights between 5 and 8:59 PM left under 15 minutes late. Southwest flies 98% of Love Field’s flights and 94% of Midway’s, and it had the lowest evening score of any airline, 54.6%. Its turns are short, with a median of 45 scheduled minutes on the ground against 60 to 72 at American, Delta and United, so a slip early in the day has little room to recover. That’s my reading: the records show the numbers and don’t show the reason.

Dallas-Fort Worth has the weakest first-flight score on this list, at 85.6%, so the morning flight isn’t a guarantee there.

The month matters as well. First flights left on time 91.2% of the time in September, and 85.1% in December, with July and March close behind it. Sunday and Monday are the weakest days, at about 86.5%, and Wednesday is the best at 89.9%.

The catch

An early flight has a catch that doesn’t show up in these tables. The 5 and 6 AM hours are when security is at its worst, so you may give back in the line what you gained at the gate. It’s still the right trade for most people, since you’re choosing a slower morning over a delay that could run into hours.

If you have to fly late in the day, look at where your plane is a couple of hours ahead. By then the day’s delays are already on the board, one way or another.

How I counted

“First flight” means the aircraft’s first scheduled departure of the local calendar date, by tail number. A plane that arrived on a red-eye and leaves again that morning would count, which is rare enough not to matter: requiring at least eight scheduled hours since the previous departure moves the figure from 88.1% to 87.9%. “On time” here means leaving the gate under 15 minutes after schedule, which is the DOT’s usual line, applied to departures. Cancelled and diverted flights are out of the on-time columns.

These are domestic flights by the large US airlines and their regional partners, September 2025 to August 2026. It’s one year, with its own weather. It says what happened to flights in general, and it can’t tell you about a specific flight on a specific day. Schedule changes also shift which flights are first.

What sits behind those late flights is in the five delay reasons.

If you’re on a later flight, Skycap shows where your plane is coming from. When it can tell why the flight is late, it says so in one line. How delays work in the app

The five delay reasons airlines report, and which they control

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

US airlines report every flight that arrives 15 minutes or more late under five causes: air carrier, extreme weather, national aviation system, late aircraft and security. Late aircraft is the largest, at 40% of delay minutes in the 12 months to August 2026. Airlines’ own problems are next at 33%.

Only one of those, air carrier, is what the DOT calls within the airline’s control, and that word is about to change meaning. A final rule published September 3, 2026 moves ten kinds of event out of the air carrier category. It was due October 19, 2026, and the DOT has since pushed it to February 1, 2027. Some articles still give the old date.

The five, and what they’re made of

Here’s each cause’s share of reported delay minutes, from public flight records for about 7 million US domestic flights, September 2025 to August 2026. The categories are the DOT’s. The arithmetic is mine.

Cause	Share of delay minutes	Largest cause on	In plain terms
Late aircraft	39.8%	39.6% of late flights	The same plane’s last flight arrived late
Air carrier	32.9%	30.1%	Crew, maintenance, cleaning, bags, fueling
National aviation system	21.1%	26.4%	Ordinary weather, traffic volume, air traffic control, airport operations
Extreme weather	6.0%	3.7%	Weather that stops flights, such as a blizzard or tornado
Security	0.1%	0.2%	Evacuations, re-screening, broken scanners

About 23% of flights that weren’t cancelled or diverted arrived 15 or more minutes late. Those flights account for 1.9 million hours of delay over the year.

Cancellations look different, and late aircraft isn’t a category there. Weather causes 61% of them, airlines 22%, the aviation system 17%. Weather that closes a runway cancels flights. Airline problems more often delay them.

The list under “air carrier” is longer than you’d guess

BTS’s technical directive gives airlines a long list of what goes here, and the Federal Register quotes it in full. Beside maintenance and late crew, it includes “slow boarding or seating”, “stowing carry-on baggage”, “last minute passenger”, “gate congestion” and “awaiting gate space”. So the passenger who took six minutes to find a bin can count as the airline’s fault, in the reports at least.

Security has its own surprise. BTS includes long lines of more than 29 minutes at screening.

Weather is bigger than 6%

The extreme weather row undersells weather, and BTS says so. Bad-but-flyable weather sits in the aviation system category: BTS says 45.8% of those delays were weather in 2020, the year its page uses. Some of the late aircraft row is weather too, passed down the line by an earlier flight, and airlines don’t report what made the inbound late.

You can’t rebuild a true total from the five numbers. If someone tells you weather causes 6% of delays, the honest reply is that it causes at least that.

Who decides

The airline does. Each one assigns the cause for its own flights, and BTS’s definition of extreme weather turns on “the judgment of the carrier”.

It matters because the DOT’s public dashboard is built on the word “controllable”, and it lists what each airline promises when a delay is its own doing. The dashboard calls a controllable delay “essentially” one caused by the airline, and gives maintenance, crew problems, cabin cleaning, baggage loading and fueling as examples. The reporting code and the dashboard are separate things, but they use the same idea, and that makes a borderline case worth arguing about. What US airlines owe you has the commitments.

What changes on February 1, 2027

Section 511(b) of the FAA Reauthorization Act of 2024 says ten kinds of event “shall not be included” in the air carrier code. The DOT’s rule puts them in a new category of their own, called Section 511(b). They are:

Aircraft cleaning after a passenger’s death
Aircraft damage from extreme weather, foreign object debris or sabotage
A bag or cargo delay from an outage in a system the airline doesn’t control
Cybersecurity attacks, if the airline met the applicable rules
An unexpected shutdown or failure of a government system
Overheated brakes after an emergency procedure
Unscheduled maintenance that shows up outside the scheduled program and can’t wait
A medical emergency through no fault of the airline
Removal of an unruly passenger
An airport closure for volcanic ash, wind or wind shear

Unscheduled maintenance is the one to watch, since maintenance is the example everyone gives of an airline’s fault. Airlines asked for the delay to February 2027, so that reporting under the new categories starts with January 1 flights. The DOT agreed.

From early 2027, expect the air carrier share to fall. Flights won’t have got better. Some delays will have moved to a new label. If you compare years, compare with care. A second change is on the way too: Section 511(a) requires a new category for delays from air traffic control instructions, and the DOT says it will handle that in future rulemaking.

The reports have gaps

The percentages are shares of delay minutes reported by the airlines for US domestic flights by large carriers and their regional partners. A late flight can list several causes, and only flights 15 or more minutes late get causes at all. The rule’s practical effect depends on how airlines code things, and I haven’t seen any 2027 data.

The most useful of the five, for a passenger, is the top one: late aircraft can be checked from your phone. For how delays pass from flight to flight, see does the first flight of the day leave on time. Trackers and airlines also disagree on the times themselves: why the tracker and the airline disagree.

When Skycap can tell why your flight is late, it says so in one line, such as low clouds at the airport you’re landing at, or the plane still on its way in. When it can’t tell, it doesn’t guess. How delays work in the app

Ground stop vs ground delay program

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

A ground stop holds flights headed to one airport at their departure points until further notice, and no flight gets a release time. A ground delay program gives each flight bound for that airport its own release time. The first is usually short and comes with little warning. The second can last all day, and it tells you how long your delay should be.

If you’re in the terminal, the difference comes down to this: in a ground stop you stay near the gate, because it can end at any moment. In a ground delay program you have a number, and you can plan around it, while still keeping an eye on it.

	Ground stop	Ground delay program
What it does	Holds flights to one airport	Spaces flights to one airport
Your release time	None until it’s lifted	A time for your flight, the EDCT
How long	A short period, in the FAA’s words	A lengthy period
Usual cause	A short-lived problem, or a pause while a longer plan is built	Too little arrival capacity for too long, often weather
Who’s affected	Flights to that airport, from listed regions	Flights to that airport, from listed regions

The FAA’s glossary gives three common reasons for a ground stop: a short spell of too much traffic, a pause to set up something longer such as a ground delay program, or an airport that can’t take arrivals at all. For a ground delay program, the most common reason it gives is low ceilings and visibility.

Both apply to the destination. A ground stop at Boston does nothing to a flight from Boston to Chicago. It holds the flight from Washington to Boston at Washington, even if the sky in Washington is clear, and that’s what confuses people at the gate.

Three advisories from one Thursday

The FAA’s command center publishes every one of these as a plain-text advisory, and Thursday, November 5, 2026 had a good spread.

Houston Hobby had a ground stop from 1:28 PM to 2:30 PM Central, for flights from the Houston and Fort Worth centers. The advisory expected an average delay of 30 minutes, a maximum of 47, a medium chance of extension, and gave the reason as “other / emergency”. It was cancelled at 1:53 PM, 25 minutes in. Anyone who wandered off for a long lunch would have cut it close.

Boston had one from 6:02 PM to 7:15 PM Eastern, because of volume. Arrivals were already holding in the air during runway construction, and the comment says the stop was “to make room for recovery” in the next hour. Maximum delay 68 minutes, average 34, medium chance of extension. A ground stop can exist to clear a line in the sky, whatever the weather at your end.

San Diego had a ground delay program from 8 AM to 8:59 PM Pacific, covering flights from every US air traffic center and 11 Canadian airports. The arrival rate was set at 24 an hour, rising to 26 for the last two hours. Maximum delay 84 minutes, average 38. The reason was volume, and the comments note arrivals and departures both on runway 27, San Diego’s only runway. Here’s part of it:

CTL ELEMENT: SAN
ARRIVALS ESTIMATED FOR: 05/1600Z - 06/0459Z
PROGRAM RATE: 24/24/24/24/24/24/24/24/24/24/24/26/26
MAXIMUM DELAY: 84
AVERAGE DELAY: 38
IMPACTING CONDITION: VOLUME / VOLUME

The times are UTC, so subtract 8 hours for Pacific in November. Each number in the rate is one hour’s arrivals.

What an EDCT means for you

EDCT stands for Expect Departure Clearance Time. It’s the time your flight should leave the runway, worked back from an arrival slot at the far end. FAA training material says controllers aim to get a flight off within 5 minutes either side.

Three things follow from that.

An EDCT is a runway time, so the door closes well before it, with taxi time added. If your EDCT is 4:10, you won’t be boarding at 4:05. The airline usually decides when to board to fit it, and the time on the board often lags.

It can also move. Airlines may swap slots among their own flights, so your flight could go earlier or later. On July 19, 2024, a ground delay program for Minneapolis was cancelled at the airline’s request, and the advisory told crews to disregard their times.

The holding happens on the ground on purpose. That’s the point of the program: the FAA’s order says it’s there to limit airborne holding. You spend the delay at a gate with a bathroom rather than circling over Lake Michigan.

What to do at the gate

If it’s a ground stop, stay close. The end of one rarely comes with notice, and a flight can board fast once it’s released. Houston’s lasted 25 minutes of a planned 62.

If it’s a ground delay program, ask the gate agent for your EDCT, or see if your airline’s app shows it. You may have time for a meal. Stay close enough to get back, since your time can move earlier.

Neither one is a reason to stay home if you’re not at the airport yet. Airlines can still board on the original schedule, and a posted delay isn’t a new departure time.

Where to see it yourself

FAA NAS Status shows current ground stops, ground delay programs and airport closures on a map. It’s public and needs no account. The advisory database has the raw text, and once you’ve read one or two it’s quicker than any news story. Flight tracking words covers the abbreviations.

These programs are one reason a flight can sit late with nothing wrong with it or its airport. A plane that arrives late because of one carries the delay onward, which is where late aircraft picks up, and Why it’s late has the rest.

An advisory is a forecast

The numbers in an advisory are forecasts by traffic managers at the time they’re written. They revise them, and the delays you see may differ. The three cases here are one day’s examples, and I haven’t counted a year of them. The public pages give the program in force, and rarely say why your particular flight got the time it did.

When air traffic control is slowing flights at your airport or at the one you’re landing at, Skycap says so on your flight with the reason it can see. It won’t tell you it’s safe to leave later, because the airline can still board on time. How delays work in the app

"Late aircraft": your plane is still somewhere else

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

“Late aircraft” means the plane assigned to your flight hasn’t got to your gate on time, so your flight can’t leave on time either. It’s the biggest single reason US flights are late: 40% of reported delay minutes in the 12 months to August 2026, ahead of the airlines’ own problems at 33%.

It’s also the most useful delay reason you can be given, because you can check it yourself. Look up where your plane is. Once it’s running late, your flight usually leaves about 45 minutes after that plane parks, and that estimate is often better than the “on time” still showing on the board.

The 45-minute rule

I followed each aircraft from one flight to its next, using public flight records for about 7 million US domestic flights over those 12 months. That gives 5.1 million pairs: an inbound flight and the flight it turned into.

The useful measure is how long before your scheduled departure the inbound reached the gate. Here’s how the next flight did.

Inbound parked	Median departure	Left under 15 min late
60+ min before your departure	4 min early	91%
50 to 60 min before	3 min early	89%
40 to 50 min before	On time	82%
30 to 40 min before	10 min late	62%
20 to 30 min before	21 min late	35%
10 to 20 min before	31 min late	11%
Under 10 min before	42 min late	1%
Up to 15 min after	54 min late	Almost none

Below about 40 minutes, the schedule stops mattering and the plane sets the time. In every late row, the aircraft sat at the gate for a median of 46 minutes before it left, whether it arrived 10 minutes short of your departure or an hour after it.

That number depends on the plane.

Aircraft and airline	Median time at the gate, when already late
Regional jets (SkyWest, Envoy, PSA, Republic)	34 min
Southwest	47 min
Alaska	50 min
American	54 min
United	56 min
Delta, JetBlue	58 to 59 min

Distance counts too. When the next flight is under 750 miles, the median is 41 minutes. Over 1,500 miles it’s 58 to 67, likely because a long flight needs more fuel, catering and cleaning.

So here’s the sum. Say you’re booked on a 3:00 PM Delta flight and its plane is due at the gate at 2:40. Add about 55 minutes, and you get a departure around 3:35, whatever the board says at 2:00.

How to find your plane

Your flight number won’t tell you. You want the aircraft.

The airline’s app or website often shows the inbound flight on the flight status page, or a “where’s my plane” link. If it doesn’t, a flight tracker can show the tail number on your flight, and looking up the tail number shows where that aircraft is now. Then look up that flight’s arrival.

Do this a couple of hours before departure. The earlier you check, the more likely the plane changes.

When to stop trusting “on time”

Airlines update their boards when they’ve decided something, and the plane’s position is often the first sign. The DOT requires them to post a status change within 30 minutes of knowing it, so the board can lag behind what you can see for yourself.

My rule for the gate: if the inbound won’t be parked 40 minutes before your departure, assume you’re leaving late. In the table, that’s the line where the chance of leaving under 15 minutes late drops from four in five to about three in five, and it falls fast from there. About 15% of all flight pairs fall under the 30-minute line. It’s common.

The rule for your couch is different. Don’t change when you leave for the airport because of it. An airline can swap to another aircraft, and flights recover: a ten-minute delay doesn’t move your leave time, and neither does this.

It gets worse through the day

Late aircraft barely exists first thing. Flights before 9 AM have 18% of their delay minutes from a late inbound, because most of those planes parked overnight. By 5 PM to 8:59 PM, it’s 48%. The day’s small delays have been stacking up behind each other for 12 hours.

That’s the main reason the first flight of the day leaves on time more often.

It also means the reason you’re told isn’t always the first cause. Airlines don’t report what made the inbound late, and BTS notes that some of this category is really weather further back up the chain. The five delay reasons goes through what each one hides.

What the records can’t show

An aircraft is only one thing that has to arrive. Crews, a gate that’s occupied and a late bag load can all hold a flight after the plane is there, and the public records don’t show who the crew was or where they came from. A swap is also invisible until it happens: Delta’s contract, like others, says it may substitute aircraft.

The times here are gate times. The pairs cover domestic flights by the large US airlines and their regional partners, only where the same aircraft flew the next leg within six hours, and they leave out cancelled and diverted flights.

If your plane is late, boarding may start late too.

Skycap shows where your plane is before you board, and tells you when its arrival looks tight for your departure. If it lands late, your boarding time moves with it. How the Flight page works

A ten-minute delay doesn't move your leave time

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

Leave at your original time. A posted delay moves an estimate, and the airline can still board on the original schedule. When the plane arriving for your flight ran 10 to 19 minutes late, 27% of the next flights still left at or before the scheduled time, and 38% left within 5 minutes of it.

The delay on the board is a guess at the time it was typed. The check-in and bag cutoffs still count from the scheduled time. And a wrong call costs far more in one direction than the other: arrive 20 minutes early and you read a book, arrive 20 minutes late and you may have missed it.

What happens to a slightly late plane

I followed each aircraft from one flight to its next through public flight records, for about 7 million US domestic flights over the 12 months to August 2026. The inbound’s lateness is the closest public stand-in for the small delays people see posted, since records can’t show what a board said at any given moment.

Inbound arrived	Median departure of the next flight	Left at or before schedule	Left within 5 min
1 to 9 min late	1 min late	48%	63%
10 to 19 min late	10 min late	27%	38%
20 to 29 min late	19 min late	17%	23%
30 to 44 min late	31 min late	12%	16%

A 20-minute delay still left about one flight in six on the original time. Those are mostly flights with room in the turn, which brings up the number that decides this: how long the plane was scheduled to sit on the ground.

Here’s a plane 10 to 19 minutes late, by its scheduled time at the gate.

Scheduled time at the gate	Median departure	Left at or before schedule	Left within 5 min
Under 45 min	18 min late	4%	12%
45 to 59 min	12 min late	17%	29%
60 to 89 min	4 min late	38%	53%
90 min or more	3 min early	68%	78%

With a long turn, a small delay disappears. That’s why the board can read “10 minutes late” and the flight can board on time anyway.

The deadlines don’t float

Cutoffs follow the scheduled departure unless the airline says otherwise. JetBlue’s page puts it directly: miss the international check-in deadline and you may be denied boarding, even if the flight is delayed. Southwest’s contract allows it to depart early when everyone who met the check-in rules is on board. Delta’s says it won’t hold a flight for passengers who aren’t at the gate on time.

The DOT also warns about the opposite problem. It calls it a “creeping delay”: the estimate gets longer as the morning goes on, because weather that should have improved didn’t, or a repair turned out to be harder. It says early estimates are sometimes hard to make. Either way, the first number you see isn’t the last.

When you can change your plans

Moving your leave time makes sense in a narrow set of cases, and all of them involve evidence stronger than a number on a board.

The flight is cancelled. Don’t go, and deal with the rebooking from wherever you are.

The airline has told you, in plain words, to come later. A new estimate on a screen isn’t that. A message saying you can arrive later is, and even then keep the original bag cutoff in mind unless it says otherwise.

The plane is a long way off. If the aircraft for your 2 PM flight hasn’t left Denver, and it’s two hours from you, then physics has a say. Late aircraft shows how to find it, and that a flight usually leaves about 45 minutes after its plane parks. Even then I’d keep some margin, since an airline can swap in a different aircraft that’s already there. That would put you back on the original time.

Everything else, go. If it’s a ground stop for your destination, it can end with no notice. A 10-minute delay on a flight with an hour-long turn is noise.

The airport isn’t a bad place to be wrong. You’d have some time near the gate, and leaving on schedule means your margin survives if security is slow too.

An inbound isn’t a posted delay

A flight can be posted late for reasons that have nothing to do with its plane: crew, a gate, weather where you’re going. Those aren’t in these tables, and I can’t show how often a posted delay held. What I can show is how much a small delay on the aircraft side carries through. Airlines may also handle one flight differently from another. Records are gate times for domestic flights, by large US airlines and their regional partners, where one aircraft flew both legs within six hours, and they leave out cancelled and diverted flights.

The airline’s app outranks all of this. Check it before you change anything.

When your flight runs late, Skycap says by how much, and the leave-by stays on the airline’s schedule. It tells you the airline can still board on the original time, so check its app before you change plans. How the leave-by works

What US airlines owe you when a flight is delayed or cancelled

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

Under US rules, an airline that cancels your flight or significantly delays it owes you a refund if you decide not to travel. That applies whatever the reason, weather included. Federal rules don’t require cash compensation, meals or a hotel for a delay. Those come from each airline’s own promises, and the DOT holds airlines to what they’ve promised.

Much of the confusion comes from pages written by companies that file claims for a cut, and from European rules that don’t apply to a US domestic flight.

This describes US rules as of November 7, 2026. It isn’t legal advice; the airline’s contract of carriage and DOT’s page are the authority.

The refund rule

The rule is 14 CFR Part 260, from a final rule published April 26, 2024 and effective June 25, 2024. The DOT’s page summarises it plainly. You’re entitled to a refund if the airline cancels your flight, or significantly delays or changes it, and you choose not to travel or accept what’s offered instead.

“Significantly” has a definition.

Change	Domestic	International
Arrives later than scheduled	3 hours or more	6 hours or more
Departs earlier than scheduled	3 hours or more	6 hours or more
Different origin or destination airport	Counts	Counts
More connections than you bought	Counts	Counts
Involuntary downgrade	Counts	Counts

There are further cases for passengers with disabilities, on the DOT page.

The refund is the full fare, with taxes and fees. It goes back in the original form of payment. The rule’s deadline for a “prompt refund” is 7 business days for a credit card purchase and 20 calendar days for anything else.

Taking the flight ends it. The DOT says that if you chose to take a significantly delayed flight or an alternative the airline offered, you aren’t entitled to a refund under its rules. Think before you accept a rebooking, and read what you’re being offered. A voucher offered in place of a refund has to come with a statement that you can have the refund.

One live exception: a flight that only got a new number counts as cancelled under the rule’s text, but the DOT has paused enforcement of that case until July 7, 2027, as long as you’re rebooked on it and it isn’t otherwise significantly changed.

What the airlines promise

Federal rules make each large airline publish a customer service plan, and the DOT keeps a dashboard of what those plans commit to when a delay or cancellation is within the airline’s control. The DOT calls these promises enforceable: it says airlines are required to adhere to them. This is the table for controllable disruptions, as shown on November 7, 2026. The page was last updated September 11, 2025.

If the airline caused it	Who promises it
Rebooking on the same airline, no charge	All nine listed
Rebooking on a partner, no charge	Alaska, American, Delta, Hawaiian, JetBlue, United
Meal or voucher, for a wait of 3+ hours	All nine
Hotel, overnight	All but Frontier
Transport to and from the hotel	All but Frontier
Credit or voucher, for a wait of 3+ hours	Alaska, Hawaiian, JetBlue, Southwest
Miles, for a wait of 3+ hours	Alaska, Hawaiian
Cash compensation	None

Those nine are Alaska, Allegiant, American, Delta, Frontier, Hawaiian, JetBlue, Southwest and United.

Read an airline’s contract as well, since the wording can be tighter than a tick mark. Delta’s contract gives its overnight hotel voucher for a disruption of more than 4 hours and leaves out weather and other force majeure events. American’s says that if a delay is beyond its control, you pay for your own hotel and meals.

When the cause is weather, the promises mostly stop. The DOT’s advice is to ask airline staff what they’ll cover.

The five delay reasons explains how “controllable” gets decided, and how a rule due on February 1, 2027 redraws that line.

Stuck on the tarmac

14 CFR 259.4 sets separate limits for a plane waiting on the ground with you aboard.

The airline must give you a chance to get off before 3 hours on a domestic flight, or 4 on an international one.
Food and water are due no later than 2 hours in.
Lavatories must keep working, and medical attention must be available.
You must be told the status once the delay passes 30 minutes.

There are exceptions: if the pilot decides that returning to the gate would risk safety or security, or if air traffic control advises it would significantly disrupt the airport. The clock on a departing flight starts when the door closes.

What was proposed, and isn’t

In December 2024 the DOT published an advance notice saying it was considering requiring cash compensation, rebooking at no charge, meals and lodging for significant disruptions. On November 17, 2025 it withdrew that notice. So there’s no pending federal requirement for cash compensation, and any article saying it’s coming is out of date.

Separately, a rule published April 24, 2026 requires airlines to post a one-page summary of passenger rights on their websites. It took effect May 26, 2026, but posting isn’t required until the DOT announces that the paperwork approval is finished.

What’s left out here

Everything above is about US rules on flights to, from or within the United States. A flight from the EU or UK can fall under their rules, which this page doesn’t cover. Being bumped from an oversold flight has its own rules. Delayed bags have one more: the DOT says the fee comes back if a domestic bag is more than 12 hours late, and how long checked bags take covers the normal case.

If you think an airline hasn’t kept a commitment, the DOT’s advice is to go to the airline first and then file a complaint with DOT.

If your flight is only running a little late, you don’t have a claim yet, and you should still leave on time.

If your flight is cancelled, Skycap says so on every screen, drops the leave-by and points you to the airline’s app. Add the new flight and it picks up from there. How delays work in the app

How long a 45-minute connection really is

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

A 45-minute connection is about 30 usable minutes, because the door closes roughly 15 minutes before departure. At big US hubs that was enough for a change within the same concourse in about 88% of the flight pairs I checked, and for a change to a far concourse in about 79%. In the evening, those numbers drop to 83% and 72%.

So 45 minutes is fine in the morning, on one ticket, with a later flight behind it if things go wrong. I wouldn’t book it on the last flight of the day.

What 45 minutes contains

The clock on your itinerary runs from scheduled arrival to scheduled departure. Three things take bites out of it.

The gate cutoff is about 15 minutes. Delta and United both say to be at the gate 15 minutes before departure.

Getting off takes longer than it feels like it should. From row 25 of a 737 or A320 at a typical load, the estimate is about 8 to 11 minutes after the door opens. From row 5, it’s about 2 to 4. Those come from a model fit to timed flights, so treat them as ranges.

Then there’s the walk. Two gates in one concourse can be 5 minutes apart. Atlanta is the classic stretch: from public map data, a Concourse E gate is about 17 to 29 minutes from the terminal curb, train included, and a Concourse T gate is about 6 to 11. Going from T to E therefore costs something like 11 to 18 minutes. That’s an estimate from walking speeds and maps, and nobody timed it with a stopwatch.

Add it up and you get two working numbers.

Change of plane	Door	Off the plane	Walk	You need
Same concourse, mid-cabin	15 min	About 8 min	About 2 to 5 min	About 25 min
Far concourse, back of the plane	15 min	About 10 min	About 15 min	About 40 min

Twenty-five and forty are my rounding, and they’re the two thresholds used below.

What actually happens to the 45 minutes

Schedules have padding, so a flight usually reaches the gate early, and the next one often leaves a little late. A 45-minute connection therefore tends to run longer than 45 minutes. The trouble is the tail.

Using public flight records, I paired every arrival and departure by one airline at each of 19 airline hubs, on the same day, with a scheduled gap of about 45 minutes, across the 12 months to August 2026. That makes about 5.2 million possible connections. For each I measured the real time from gate arrival to gate departure.

Scheduled gap	Median real gap	Under 25 min	Under 40 min
45 min	56 min	12%	21%
60 min	70 min	8%	12%
90 min	101 min	5%	6%
120 min	132 min	3%	4%

Going from 45 minutes to 60 nearly halves the far-concourse failures. Going from 90 to 120 buys much less. The first 15 extra minutes do the most.

The hour matters as much as the airport. Here are the 45-minute pairs by when the first flight was due in.

Scheduled arrival	Under 25 min	Under 40 min
Before noon	8%	17%
Noon to 4:59 PM	11%	20%
5 PM or later	17%	28%

Airports differ too.

Airline and hub	Under 25 min	Under 40 min
Southwest, Baltimore	8%	19%
Delta, Atlanta	10%	19%
United, Denver	11%	20%
American, Charlotte	12%	20%
American, Dallas-Fort Worth	14%	22%
United, O’Hare	14%	27%
United, San Francisco	22%	36%

Atlanta’s reputation for size doesn’t show up here, because Delta’s flights there tend to arrive early: 74% of the inbound flights in those pairs reached the gate ahead of schedule. San Francisco stands out, and these records can’t say why.

One local detail: a tracker’s “landed” isn’t the gate. Taxi-in at O’Hare has a median of 18 minutes and a 90th percentile of 36, and at the 7 PM hour those rise to 23 and 52. The real-gap numbers above already include taxi time, but your own watch doesn’t. Sitting on the ramp at O’Hare with 20 minutes left isn’t yet a missed connection. The taxi-in table has the figure for your hub.

What this can’t show

These are possible connections, not real passengers. The records don’t show who was on board, whether an airline held a flight for connecting passengers, or whether a gate agent reopened a door. Cancelled flights are left out, which flatters every row. Only flights with the same operating airline are paired, so regional feeders aren’t included, and domestic flights only.

International arrivals are a separate case. Coming into the US from abroad, you clear passport control, collect your bag, recheck it and go through security again. I’d want 90 minutes there at the least, and that’s judgement, since nobody publishes the chain.

Two things worth doing

Book the gap, then pick the seat. If a 45-minute connection is the only option, a seat near the front is worth more than a window: row 25 costs you about 6 minutes against row 5.

Keep it on one ticket. Airlines don’t promise that you’ll make a connection (Delta’s and Southwest’s contracts both say so), but Delta’s rule on missed connections, for one, rebooks you or refunds the unused part. On two tickets nobody has checked the gap, and the rebooking is between you and the second airline. Minimum connection time explains what the booking system was checking when it sold you the gap, boarding time vs departure time covers how close to departure the door really closes, and how long it takes to get off the plane has the row-by-row detail.

Skycap keeps both flights as one trip and shows the minutes left between them. It names the gates and any change of terminal once the airline has set them, and says when the gap looks tight. How connections work

Walking estimates use public map data.

Boarding time vs departure time

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

Departure time is when the plane is due to leave the gate. Boarding usually starts 30 to 40 minutes before that, and the door closes about 15 minutes before it. So the number that governs your day sits about 15 minutes ahead of the one printed largest on your ticket.

Takeoff is later again. Over the 12 months to August 2026, the median taxi out at US airports was 16 minutes, and the 90th percentile was 30.

The gate cutoff by airline

Boarding start times are mostly unpublished and vary by aircraft. Gate cutoffs are published, in contracts and on airline sites, and they’re the ones with consequences. All of these are for domestic flights, as read on November 7, 2026.

Airline	The rule	What it says happens
Delta	At the gate, ready to board, 15 min before departure	Reservation may be cancelled; Delta won’t hold the flight
American	At the gate 30 min before; boarding ends at 15	Seat may go to someone else
United	At the gate 15 min before	Reservation may be cancelled, boarding denied
Southwest	In the gate area 10 min before	Reservation may be cancelled
JetBlue	On board 15 min before	May be refused, even with a boarding pass

International flights run earlier. American wants you at the gate 45 minutes before, Delta suggests the same, and JetBlue wants you on board at 20. United’s contract says 30 at some international airports. Your own boarding pass overrides this table.

I couldn’t read a current gate rule on Alaska’s site when I checked. Its 2015 post on boarding is still online and lays out a sequence from 40 minutes down to 20, and I’d treat that as a guide to the shape of things.

Southwest is the one that says when boarding can start: as early as 30 minutes before departure. The time printed on your pass is the one to trust.

The time on your ticket can run early

Plenty of people treat departure time as a promise that the plane will be there until then. It’s closer to an estimate. In the records for about 7 million US domestic flights over the year to August 2026:

Left the gate	Share of flights
Before scheduled time	57%
5+ min early	36%
10+ min early	9%
15+ min early	1%

That’s why the cutoff matters. An airline that has everyone on board has every reason to shut the door. Southwest’s contract says outright that it may depart early once everyone who met the check-in rules is on board. If you walk up to the gate 12 minutes before the printed time, the plane may already have gone.

The reverse is true when a flight is delayed: boarding can begin later than printed. A posted delay doesn’t move the cutoff until the airline moves it. Late aircraft explains how to tell when a delay is real.

Why trackers show several times

An arrival or departure in airline reports means the gate, which is defined that way in 14 CFR 234.4. “Wheels up” and “wheels down” are the runway. A tracker that shows “departed” while you’re sitting at the end of a taxiway isn’t wrong, and neither is your airline’s app showing a different time: why they disagree goes through it.

The times also move in a connection. If your second flight is 45 minutes after the first lands, you really have about 30: how long a 45-minute connection really is has the numbers. For the hours before all that, see when to leave for the airport.

These rules move

Airlines change these rules, and they vary by airport and aircraft. The table is a dated reading of each airline’s public page. The early-departure figures cover US airlines’ domestic flights, and exclude cancelled and diverted ones.

Skycap counts down to boarding on your flight, not to departure. If your plane is arriving late, the boarding time moves with it and the page says why. How it works at the airport

Minimum connection time: who sets it and what it protects

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

Minimum connection time is the shortest gap between two flights that an airline may sell on one ticket at a given airport. It’s a booking rule, set by the airlines themselves. It doesn’t promise that you’ll make the flight, and it takes no account of whether your first flight is running late.

When a website offers you a 38-minute connection, that means the gap passed a filter. It tells you the itinerary was legal to sell on the day it was sold. That’s all it tells you.

Who sets it

The system runs through IATA, the airlines’ trade association. IATA’s user guide says minimum connecting times are governed by Resolution 765, and that ticketing outlets and automated reservations systems must observe them.

The numbers come from the airlines at each airport. A local connecting-times group or airline operating committee agrees a value and submits it to IATA. IATA’s station standard covers more than 400 of the most connected airports, by its own account.

On top of that standard, each airline can file exceptions. It can ask for less time between its own gates, or more on a route it knows is awkward. The guide’s worked examples show how many layers there are, down to exceptions by aircraft type and by season. In one example, a Delta connection at Amsterdam falls back to the airport standard of 50 minutes because no exception applies.

The DOT doesn’t set these values. I couldn’t find a federal rule that does.

What it protects

IATA defines the term as “the shortest time interval required in order to transfer a passenger and his luggage” between flights. Two words in that sentence matter.

“Required” is measured on schedules. IATA says the time is based on the scheduled arrival and departure of the two flights. So the filter ran once, against timetables, possibly months before you flew.

“Luggage” is the part people forget. The minimum covers moving you and your bag, so on an ordinary day your suitcase is meant to make the flight too.

What it protects, in practice, is the airline’s schedule. A filtered itinerary is one the airline believes it can operate on an ordinary day, and rebook on the days it can’t.

What it ignores

Your inbound delay is the big one. If your first flight is 30 minutes late, the filter doesn’t run again.

Your row doesn’t figure in it. Getting off from row 30 of a full 737 takes longer than from row 5, and the rule treats those people alike.

Nor does the hour. In a year of public flight records, about 21% of 45-minute pairs at big hubs left under 40 minutes between gate arrival and gate departure, and in the evening it was 28%. How long a 45-minute connection really is has the method and the table by hub.

The minimum is also the same for someone on the last flight home and someone with three more flights to choose from that evening.

Two tickets are a different thing

Everything above applies to one ticket. If you buy two separate tickets, nothing checks the gap. American’s agency guidance says as much: on separate tickets, the minimum connecting time usually won’t be enough. You may also need to collect your bag and go back through security, and rebooking a missed second flight is between you and that airline.

My rule is three hours between separate tickets at a US airport, and I’d spend the night before anything I can’t move. That’s opinion, and a cautious one.

Why it’s hard to look up

The tables are sold to airlines and booking systems, so they’re rarely printed where a passenger can read them. Airline contracts don’t help: Delta’s says only that it isn’t responsible for making connections, and Southwest’s says it doesn’t guarantee them. The practical test is the one the system uses. If an airline’s own site will sell you the itinerary on one ticket, the gap meets the minimum. That’s the floor, and you get to decide whether to book something higher.

A late inbound is what breaks most tight connections, and late aircraft covers how to spot one early. The rest of the trip to the gate is in Leaving for the airport.

Once you’ve saved both flights, Skycap shows the real gap as it changes, using the latest time for each. If the first flight runs late, the number shrinks in front of you and it says so. How connections work

What PreCheck actually saves, and when it doesn't

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

PreCheck saves up to about 35 minutes at a big airport’s worst hour, about 16 minutes on an average daytime hour at a big airport, and close to nothing at a small one. It does nothing for bag drop, the walk to your gate or the gate cutoff. At $85 or less for five years, it pays for itself if you leave from a busy airport a few times a year, and I wouldn’t bother for one trip a year from a small field.

Many reviews of PreCheck are there to sell you a credit card that pays for it, so they quote the best case. Here’s the arithmetic for the ordinary one.

The saving depends on the hour

The TSA says about 99% of PreCheck passengers wait less than 10 minutes. I can’t test that claim: the public counts I use don’t split out PreCheck lanes. If you take it at face value, then your saving is the gap between the top of the usual band and 10 minutes.

Usual band for that hour	Allow without PreCheck	Allow with PreCheck	Saving, at most
Under 10 min	10 min	10 min	0 min
10 to 20 min	20 min	10 min	10 min
20 to 30 min	30 min	10 min	20 min
30+ min	45 min	10 min	35 min

That’s a planning calculation. On a day the line runs short, you save less.

Then the question becomes how often you’re in each row. Across the 30 airports with the busiest checkpoints, from 5 AM to 8:59 PM, 59% of airport-hours are in the top two bands. Across all 426 airports, it’s 4%. Averaged over those daytime hours, the top-of-band allowance comes to about 26 minutes at the big 30 and about 14 across the country, so the typical planning saving is about 16 minutes at a big airport and 4 at the average one.

Take a Thursday at 6 AM at O’Hare, where the usual band is 30+. PreCheck could plausibly take 35 minutes off your morning. The same flight from Burbank, where the band never goes above 10 to 20, saves you 10 at the very most. The worst hours table shows where your airport sits.

What it can’t touch

In the chain of steps before a flight, security is one link. If you check a bag, the airline’s cutoff is still 45 to 60 minutes before departure, and PreCheck doesn’t move it. The walk to Concourse E at Atlanta is the same length. The gate cutoff stays where it is.

So the person who gains most is someone with no checked bag, leaving at a busy hour. With a bag at 6 AM, you still have to clear the counter, and if that line is long, the time you saved at security goes to waiting there instead.

It doesn’t help after you land, either. Coming home from abroad, you go through passport control, and getting from the gate to the curb is untouched.

The shoe change narrowed the gap

Keeping your shoes on used to be half the appeal. DHS ended the shoes-off rule at domestic airports in July 2025, so everyone now keeps them on. PreCheck still lets you leave electronics and 3-1-1 liquids in your bag, and keep your belt and light jacket on. That’s a smaller difference than it was, and a shorter line is now most of what you’re buying.

The release promised shorter waits for everyone, and didn’t give a measurement. If standard lanes got faster, PreCheck’s saving shrank by the same amount.

The cost

As of November 7, 2026, a new five-year membership costs $79.75 to $85, depending on which enrollment provider you use, and online renewal starts at $58.75. Global Entry costs $120, also for five years, and includes PreCheck. If you leave the country even once every couple of years, it’s the better buy, given the extra $40 and the faster way back in.

Three small print items. Your Known Traveler Number has to be in the airline reservation, and the PreCheck indicator has to show on your boarding pass, or the lane won’t take you. The TSA says no individual is guaranteed expedited screening. And its FAQ notes that a PreCheck lane isn’t always available, in which case you may be sent to a standard lane.

None of this measures a PreCheck lane

The bands describe an airport as a whole, on a normal week, and the table leaves out holiday weeks, when the difference may be larger. Airport-hours aren’t passengers either: a busy hour holds far more people than a quiet one, so passengers meet the top bands more often than the percentages suggest.

Typical security times are estimated from public passenger-throughput data. Not affiliated with any airport, airline or government agency.

Skycap doesn’t know whether you have PreCheck, so the leave-by uses the time you set for being at the airport. Pick a shorter one in “Be at the airport” for this flight or in Settings for all of them, and the leave-by follows. How the leave-by works

When to actually leave for the airport

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

For a US domestic flight, be at the airport between 90 minutes and 2 hours before departure, then add the drive. Use 90 minutes at a small airport or in a quiet hour. Use 2 hours at a big hub at a busy hour. For an international flight, add an hour to either.

American and Delta both recommend 2 hours domestic and 3 international, as of November 2026. That’s a fair number for someone the airline knows nothing about. You know your airport, your hour and whether you have a bag, so you can do better than a default.

The way to do it is to count backward from the one deadline that can’t move.

Count back from the gate

The deadline is the gate cutoff, usually 15 minutes before departure. Everything else stacks up behind it.

Step	Time to allow	How firm is it
Gate cutoff	10 to 30 min before departure	The airline’s rule
Walk to the gate	3 to 25 min	Depends on airport and concourse
Security	10, 20, 30 or 45 min	The top of the usual band for that hour
Bag drop	Done 45 to 60 min before departure	The airline’s rule; the line isn’t measured
Parking, train, rental return	Yours to estimate	Often the one people forget
The drive	Whatever the map says now	Check it when you leave, not the night before

I roll the gate cutoff, the walk and the bag drop into one allowance: 45 minutes at a small airport, 60 at a mid-size one, 75 at a big hub, and an extra 60 for an international flight. That allowance is judgement, and nobody publishes a measured version of it. Then I add the top of the usual security band for the hour you’ll arrive, and never go below 90 minutes, because the band describes a normal day and not every day is.

Here’s what that gives.

Flight	Security at that hour	Be at the airport
O’Hare, Thursday, 8:30 AM	30+ min	2 hr before
O’Hare, Tuesday, 8:30 AM	20 to 30 min	1 hr 45 before
Las Vegas, Sunday, noon	30+ min	2 hr before
JFK, Thursday, 6 PM, domestic	30+ min	2 hr before
Denver, Tuesday, 2 PM	10 to 20 min	1 hr 35 before
Midway, Tuesday, 1 PM	10 to 20 min	1 hr 30 before
Burbank, Thursday, 7 AM	10 to 20 min	1 hr 30 before
JFK, Thursday, 8 PM, international	30+ min	3 hr before
Atlanta, Tuesday, 11 AM, international	20 to 30 min	2 hr 45 before

The two-hour rule is right for the busy rows and has no slack in them. It’s 30 minutes too long at Burbank, and that’s half an hour you could spend at home. The bands come from the security hours table, built from six months of public passenger-throughput data.

One morning, drawn out

Take a Thursday 8:30 AM flight on United from a Concourse C gate at O’Hare, with one checked bag.

Time	Where you are	Minutes
5:40 AM	Leave home, if the drive is 50 minutes	50
6:30 AM	At the curb	15 for bag drop, which is a guess
6:45 AM	Bag dropped, in the security line	45 at the top of the 30+ band
7:30 AM	Through security, walking to the gate	15, the long end for Concourse C
7:45 AM	At the gate	30 spare
8:15 AM	United’s gate cutoff	15 before departure

On a good morning, you reach the gate by about 7:20 with half an hour before boarding starts. That’s the correct outcome. You’re buying protection from the top of the band, and on most days you won’t need all of it. The same flight on a Tuesday lets you leave 15 minutes later.

The airline’s numbers

These are the deadlines that cost you the flight if you miss them. All read on November 7, 2026, for domestic flights.

Airline	Bags checked by	At the gate by
Delta	45 min (60 at JFK)	15 min
American	45 min	30 min, boarding ends at 15
United	60 min in its contract	15 min
Southwest	45 min	10 min
JetBlue	45 min	On board by 15 min

American’s check-in cutoff at the Tampa rental car center is 90 minutes, a detail that only matters once. All of these run from the scheduled departure, so treat them as fixed even when the board says the flight is late.

Where the arithmetic stops working

The table covers ordinary weeks. Memorial Day, Independence Day and Labor Day weeks aren’t in it, and neither is Thanksgiving or Christmas. For a holiday, I’d add 30 minutes and not feel clever about it. That’s a judgement call too.

The bands are estimates from public data, and they aren’t a live line. A broken scanner or a thin morning shift won’t show. PreCheck isn’t split out either: what it changes has its own page. The TSA’s advice is to check with your airline, since times vary by airport and date.

If you’ve never cut it this fine, try it first on a flight that doesn’t matter. And if the board shows a delay when you’re ready to go, go anyway. The hours to avoid are in the worst hours at airport security, and the rest of the morning is in Leaving for the airport.

Skycap works your leave-by back from the airline’s schedule, using the drive as it stands now and the time you choose to be at the airport. It shows the usual security band for that hour beside it. How the leave-by works

Which lounge can you actually get into

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

Which lounge you can use depends on your flight more than your card. The same Amex Platinum opens 11 of Atlanta’s 13 lounges on a Delta flight and 2 of O’Hare’s 14 on a United one. Priority Pass, the name most people know, opens nothing at Denver or Newark.

So ask three questions in order: which airline is operating this flight, what fare or cabin am I in, and what’s in my terminal past security. The card comes fourth.

Lounge finders list what exists. For the numbers here I read the airport, airline and lounge pages for 161 airports, up to November 6, 2026, and restated the rules by flight. Of those 161, 82 have no lounge at all. The other 79 have 422 between them.

Priority Pass opens 123 of those 422

At the big US hubs, the count is short, and where they sit matters more than how many there are.

Airport	Lounges	Priority Pass	Where they are
Denver	8	0
Newark	6	0
Los Angeles	17	1	Tom Bradley International
Atlanta	13	2	Both in Concourse F
O’Hare	14	2	Both in Terminal 5
Dallas-Fort Worth	20	2	Terminals D and E
JFK	19	8	Terminals 1 and 4

Atlanta’s two are a train ride away from most domestic gates, which is fine on a long layover and silly on a short one. O’Hare’s two are in Terminal 5, and United and American leave from Terminals 1, 2 and 3, so expect to go through security again. JFK has the longest list, and none of it is in Terminal 5, where JetBlue flies. A JetBlue passenger holding Priority Pass at JFK has eight lounges and no way to reach them without leaving security.

Some memberships cap your visits, and space isn’t guaranteed at any of them.

Amex Platinum, flight by flight

The Platinum card is the best example of a card that changes with your ticket.

Your flight	What opens	Count
Delta, Atlanta, not Basic	Sky Clubs, Centurion, Priority Pass	11 of 13
American, Dallas-Fort Worth	Centurion in D, The Club in D, Plaza Premium in E	3 of 20
United, Denver	The Centurion Lounge only	1 of 8
United, O’Hare	Two in Terminal 5	2 of 14

On Delta, three conditions sit on top. The flight must be operated by Delta and can’t be Basic Economy. You get 10 visits per Medallion Year unless you spend $75,000 on the card in a calendar year. And you can enter only within 3 hours of departure, or during a layover. A flight booked on Delta but flown by a partner may not count.

Centurion has a 3-hour limit too. Priority Pass through the card needs enrolling first, so sort that out at home and not at the door.

Status has more traps than cards

Airline status looks simple and isn’t. These are the ones that catch people.

Status	The catch
United Premier Gold	United Clubs in the US need an international Star Alliance flight. A domestic United flight doesn’t count
AAdvantage or Atmos oneworld Sapphire or Emerald	Lounge access through status needs an international flight, and North America and the Caribbean don’t count
Delta Gold, Platinum, Diamond Medallion	Sky Clubs need Delta One, Premium Select internationally, or Delta First to Mexico, Canada or Central America. Partner lounges work in any cabin

First class is a trap of its own. Domestic first on American, Delta, United and Alaska doesn’t get you into most lounges. International business does, on many airlines, though American’s rule leaves out Canada, the Caribbean and most of Mexico.

How I’d decide

If you fly one airline from one hub, a club membership or that airline’s card gets you more than a general lounge card.

If you fly a mix of airlines through big hubs, check the table above for your home airport before paying for anything. A card with a long lounge list and none in your terminal is a card you’ll stop thinking about.

If your connection is under an hour, don’t plan on a lounge at all. The train and the line at the desk will eat it, and you’ll be working out when boarding really starts over a biscuit. On a long delay it’s a different story: stay near the gate until the new time holds, then go.

These rules change often

These are facts restated from public pages on the date above, and rules change often. Hours are as listed by each source. Some American and United pages were harder to confirm than others, so check the door rule on your airline’s site before you count on it. United Polaris lounges aren’t covered, because the rule depends on a ticket detail I couldn’t source. Other airports are at the lounge access page, and the rest of the morning is in when to leave for the airport.

Tell Skycap what you carry and it lists the lounges past security that you can use on this flight, with what gets you in. A lounge it couldn’t confirm says so. How it works at the airport

The worst hours at airport security, from six months of screening counts

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

The worst hours at US airport security are 6 and 7 AM on Mondays, Thursdays, Fridays and Sundays. Tuesday is the quiet day. That pattern comes from 23 ordinary weeks between May 3 and October 31, 2026, at 426 airports, and at most of those airports it barely matters.

If you can choose your flight, choose the day first. At the 12 busiest checkpoints, nine sit in the top band at 7 AM on a Friday. Moving to 9 AM clears four of them, and moving to Tuesday clears all nine. “Avoid the morning” is poor advice at JFK and San Francisco, where the bad hours come later.

Tuesday is the quiet day

The table has one row for each hour of the week at each airport, and each hour lands in one of four bands: under 10 minutes, 10 to 20, 20 to 30, or 30 and up. This is how many airport-hours reach the top band on each weekday.

Day	Airport-hours at 30+ min
Monday	51
Tuesday	1
Wednesday	8
Thursday	52
Friday	54
Saturday	22
Sunday	52

The one Tuesday hour is Boston at 6 AM. Wednesday has eight, at Newark, JFK, Los Angeles and San Francisco. Thursday looks like Friday.

The 6 and 7 AM hours hold 71 of the 240 top-band hours. Add 5 and 8 AM and it’s 115, close to half.

Twelve airports do most of this

Twelve airports reach the top band: O’Hare, Los Angeles, Atlanta, JFK, Denver, Dallas-Fort Worth, San Francisco, Orlando, Las Vegas, Boston, Newark and Seattle. Those are also the only 12 airports whose busiest typical hour tops 5,000 people, and I cap every smaller airport below the top band, so the number 12 is partly a result of the method. Read it as “the 12 biggest checkpoints”, and don’t read it as proof that Phoenix never has a long line.

Their bad hours differ.

Airport	When it usually reaches 30+ min
Atlanta	6 to 9 AM, and a second peak from 1 to 4 PM
O’Hare	5 to 8 AM, then 1 to 5 PM, and Sunday from noon to 5 PM
Dallas-Fort Worth	6 to 10 AM, worst on Thursday
Denver	8 to 10 AM, from 6 AM on Friday
Los Angeles	7 to 11 AM
Seattle	6 to 11 AM on Monday, Thursday and Friday, 8 to 11 AM on weekends
San Francisco	7 AM, then 9 AM to 1 PM
Orlando	5 to 8 AM, worst on Sunday and Monday
Boston	5 to 8 AM, with a smaller peak from 2 to 5 PM
Newark	5 to 8 AM, then 2 to 6 PM
JFK	2 to 6 PM on most days, plus the 6 AM hour
Las Vegas	Sunday from 5 AM through the noon hour

Las Vegas on Sunday is the longest unbroken stretch in the table: eight hours in the top band. JFK’s afternoon is probably people leaving on the evening flights abroad, though the counts can’t say where anyone is going.

The quiet end of the day holds up. At these 12 airports, the 7 PM and 8 PM hours are the most likely to sit under 20 minutes, in 75 and 78 of the 84 airport-hours.

73% is the wrong number to quote

Across all 426 airports, 73% of airport-hours are “under 10 min”. That figure is true and not much use, because it counts 3 AM at Burlington, Vermont.

Take the 30 airports with the busiest checkpoints and keep only the hours from 5 AM to 8:59 PM, and it reads very differently.

Band	Share of those hours
Under 10 min	10%
10 to 20 min	31%
20 to 30 min	52%
30+ min	7%

If you fly from a big airport during the day, the middle of the table is your normal. Plan on 20 to 30 minutes and be pleased when it’s less.

Small airports are a different trip. Burbank never goes above the 10 to 20 band, and Midway and Baltimore sit in it from early morning to night. If you’re leaving from a place like that, the hour barely matters.

How the table was made

I started from public passenger-throughput data: the number of people screened at each checkpoint in each hour. For each airport, weekday and hour I took the median across 23 weeks. Three holiday weeks were left out (Memorial Day, Independence Day, Labor Day).

Each median was then divided by that airport’s 95th-percentile hour, which gives “how close to as busy as this airport gets”. Under 35% is the first band, under 65% the second, under 90% the third, and the rest the fourth. Only airports over 5,000 people in their busiest typical hour can reach 30+, and only those over 2,500 can reach 20 to 30. The cap is a judgement: a smaller airport’s peak is fewer lanes and a shorter line, and the counts can’t show how long that line was.

The security hours data page has the method in full.

What the counts can’t see

These are bands, and they describe a normal week. Nobody at the checkpoint timed anything. A count of people screened shows demand, capped by how many lanes were open. It says nothing about staffing that day, a broken scanner, or a storm that sent two banks of flights through at once.

Three limits matter when you plan around it.

Holiday weeks aren’t in the table. The Sunday after Thanksgiving is not a typical Sunday.

There’s no split for PreCheck lanes, so these bands describe the airport as a whole.

The count also trails the line. A queue that forms at 4:45 AM shows up as people screened at 5:15, so read each peak as starting a little earlier than the hour it’s listed under.

If your flight is at 7 AM on a Friday from one of the 12, add the time and stop reading. The arithmetic is in when to leave for the airport, what PreCheck changes follows from these bands, and a delay on the board doesn’t move any of this.

Skycap looks up this table for your airport at the hour you’ll get there and shows the usual band beside your leave-by time. It says “usually”, because that’s as far as the counts go. How the leave-by works

Typical security times are estimated from public passenger-throughput data. Not affiliated with any airport, airline or government agency.

Flight Live Activities on iPhone, explained

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

A Live Activity is the card that stays on your iPhone’s Lock Screen and in the Dynamic Island while something is under way. For a flight, it usually shows the next thing that happens: boarding, take-off, landing, the arrival gate. Apple gives it up to 8 hours of updates. After that, iOS ends it, and it can sit on the Lock Screen for up to 4 more hours.

It arrived in iOS 16.1, and Apple’s own example of a good use is flight updates. There’s no separate app. An app you already have starts one when it decides the moment has come.

One card that changes

A notification is a new message every time. A Live Activity is a single card that the app updates, so your Lock Screen doesn’t pile up with delay alerts, and you don’t open anything to read it.

On iPhones with a Dynamic Island, it moves there when you unlock. Touch and hold to expand it. The iPhone 18 Pro models can show up to three at once. Others show fewer, so a timer or a recording can take your flight’s place. On macOS Tahoe 26 or later it can appear on your Mac, and it can also show in CarPlay.

The 8-hour limit

A Live Activity can be active for 8 hours, starting from when it started and not from when your flight left.

Do the arithmetic for a nine-hour flight. If the app started the activity when you arrived at the airport, it runs out over the ocean, long before the part you most want to see. So on long flights, a well-behaved app starts it later, near boarding, and accepts that you’ll see less of the morning. An app can also start one remotely with a push, which needs a connection on your phone at that moment.

When the limit passes, iOS removes the card from the Dynamic Island at once. It stays on the Lock Screen, frozen at its last state, until you clear it or up to 4 hours go by.

With no signal

A Live Activity has no connection of its own. It shows what it was last given, and it gets news from the app or from a push over the network.

Two things keep working in airplane mode. A countdown runs on the phone, so “lands in 2 h 14 min” keeps counting down, based on the last estimate. Anything already on the card stays there. What can’t reach you is news: a revised landing time, a gate change or a new delay. If your flight has Wi-Fi and you’ve joined it, updates can get through, depending on what the airline’s network allows.

So read the countdown in flight as an estimate made before you lost signal. A phone that has been offline for three hours is showing a three-hour-old opinion.

Apple’s documents let an app mark content as out of date after a set time, and a good app will use that. When you land and reconnect, expect a jump.

Turning one off

Swipe left on the card to clear it. That removes this one activity. To stop an app starting them at all, go to Settings, tap Apps, tap the app, then tap Live Activities. Not all apps have that setting. There’s no single switch for every app that I could find in Apple’s guides.

If a card won’t leave after the flight, it’s usually an app that didn’t end its activity properly. It will go by itself within hours, and swiping clears it sooner.

Things this page isn’t sure about

Apple doesn’t publish a clear statement of how much a Live Activity costs in battery, and it limits how often apps can update through a budget it doesn’t spell out. Apple’s documents also change between releases, so I’ve quoted the 8 and 4 hour figures as they stood on Nov 7 2026.

Your phone may not be the only thing that knows about the flight. If someone’s meeting you, send them a link so they aren’t reading your Lock Screen by text message. A flight marked “landed” and one marked “arrived” are different events, as this page explains. Once you’re on the ground, the hour ahead is laid out in how long from landing to the curb.

On your flight, the Skycap Live Activity starts about 3 hours before the gate time, or later on a long flight so landing still falls inside Apple’s limit. It keeps counting with no signal, and you can start or remove it from the Flight page’s More menu. What it shows in the air

Flight tracking words: wheels up, in-blocks, EDCT, ETA

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

“Wheels” words are about the runway, “blocks” and “gate” words are about the parking spot, and most airline times are gate times. Where a regulator defines a term, I’ve used its meaning. Where it’s airline shorthand, I’ve said so.

The four moments of a flight

Airlines record these four and call them OOOI: out, off, on, in. Blocks are the wheel chocks.

Moment	Also called	What it means
Out	Off-blocks, pushback, gate departure	The plane leaves the gate. Your ticket’s departure time means this
Off	Wheels up, wheels-off	The plane leaves the runway. The FAA’s glossary calls this departure time, which is a trap, because your ticket doesn’t
On	Wheels down, wheels-on	The plane touches the runway. Most trackers mean this by “landed”
In	In-blocks, gate arrival	The plane parks. US reporting rules count it when the pilot sets the parking brake. Your arrival time means this

Why the tracker and the airline disagree follows from that table.

Times and estimates

Term	Meaning
Actual time	A recorded event. Anything else is an estimate
Block time	Gate to gate, out to in. Schedules are built on this, with room for taxiing
Air time	Wheels-off to wheels-on
ETA	Estimated time of arrival. For scheduled airlines, the FAA defines it as arrival at the gate. Trackers sometimes use it for landing, so check
ETE	Estimated time en route, the flying time from lift-off to touchdown
On time	In US reporting, arriving less than 15 minutes after the published arrival time. A flight 14 minutes late is on time
Taxi-in	Wheels down to the gate. National median 7 minutes, with a wide spread
Taxi-out	Gate to wheels off

Why you’re waiting

Term	Meaning
EDCT	Expect Departure Clearance Time. The runway release time given to a flight in a traffic management program. If you have one, the airline didn’t choose that delay
Ground stop	Aircraft that meet set criteria stay on the ground, for example all departures to one airport. More in Why it’s late
Ground delay program	A traffic management process that holds aircraft on the ground to limit airborne holding
Gate hold	Keeping a plane at the gate or another ground spot when departure delays run past 15 minutes
Holding	Flying a racetrack in the air until there’s room to land. How to spot one
Diverted	Required to land somewhere other than the scheduled destination, for reasons beyond the pilot’s or airline’s control

In the air

Term	Meaning
ADS-B	Automatic Dependent Surveillance-Broadcast. The plane sends its position about once a second. More here
Transponder	The radio on board that answers ground radar. It often carries ADS-B too
Squawk	A controller’s word: set the transponder to a four-digit code
Call sign	The name a flight uses on the radio. It may not match your flight number, especially on a regional carrier
Flight level	Altitude at a standard pressure setting, in hundreds of feet. FL350 is 35,000 feet on the altimeter
Step climb	Airline term for moving up a level in cruise as fuel burns off
Top of descent	The point where the plane begins its initial descent
Assumed path	Tracker term for an estimated position with no report behind it. Why that happens

On the ground afterwards

Term	Meaning
At the curb	No regulator’s term. It’s how I describe the moment you’re outside the terminal. How long that takes
Passport control	The border booth when you land in the US from abroad. Waits, by terminal

How flight trackers work, and why the position lags

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

A flight tracker is a map of other people’s radio reports. Planes broadcast their position, receivers collect it, a server merges it, and your phone asks the server what’s new. Each hop adds seconds, and over an ocean the dot may be an estimate with no report behind it at all. A good position is a few seconds old. A bad one can be minutes old and still look perfectly smooth.

Most trackers don’t tell you which you’re looking at. I think they should.

Where a position comes from

Source	What it gives a tracker	The catch
The plane’s own broadcast (ADS-B)	Position from satellite navigation, about once a second, with speed, altitude and call sign	Needs a receiver within line of sight. Hobbyists and companies both run them
Satellites listening from above	The same broadcast, over oceans and poles	Usually a paid service
Timing the signal (multilateration)	A position for planes that don’t report one, worked out from when the signal reached each receiver	Needs at least four receivers in range, so it works best around cities
Air traffic control	Flight data from radar and the same broadcasts	Rules and access differ by country
The airline	Timed messages for pushback, take-off, landing and arrival at the gate	Says when an event happened, not where the plane is
The schedule and flight plan	Route and times	An intention, not a sighting

A broadcast doesn’t say where a plane is going, so a tracker has to match the call sign to a schedule or flight plan to show a route.

Why the dot is behind the plane

Follow one report. The plane sends it. A receiver hears it almost instantly, then forwards it with others. A server merges several sources, drops duplicates and publishes. Your app asks every so many seconds, then draws it.

None of those steps is long, but they add up, and the plane doesn’t wait. At 500 mph, a jet covers about 730 feet a second.

Age of the position	How far the plane has moved at 500 mph
1 second	730 feet
5 seconds	0.7 miles
15 seconds	2.1 miles
60 seconds	8.3 miles

So when you match a dot against a plane in the sky, a 15-second-old position puts the dot visibly behind it. Some apps slide the dot forward between reports so it looks continuous. That’s fine as long as the slide is small. It’s a guess, though, and the smoothness doesn’t tell you how good it is.

Update rates also vary with the source. Broadcasts arrive every second, while other systems report far less often. A tracker that merges them can have a plane that moves in small steps over land, then in big jumps elsewhere.

The assumed path

When a plane leaves coverage, some trackers don’t stop the dot. They carry it along the filed route at the expected speed, sometimes for hours, and it looks exactly like the live part of the flight.

That’s reasonable for a flight across the Atlantic. The plane is very likely close to that line. But the dot isn’t evidence of anything. It can’t show a diversion, a hold or a return until the next real report arrives.

A few signs that you’re looking at an estimate:

What you see	What it suggests
A perfectly straight or perfectly curved line, no small wobbles	Assumed path
Altitude and speed flat for a long time, with no change at all	Numbers may be held, not reported
A jump when the plane reaches land	The real report came back
“Estimated” or an old timestamp in the details	The tracker is telling you, which is to its credit

“Live” should mean someone heard this plane moments ago. It often means the app is open.

What this means for you

If you’re watching someone’s flight over land in the US, the position is good to within a few seconds’ flying. If you’re watching over the middle of an ocean, treat the dot as the plan. Wait for landfall before you read anything into it, in either direction.

For timing, don’t rely on the dot. The landing and gate times are separate events, and they come through different routes than the position does. For a person waiting at the airport, the estimate in how long from landing to the curb matters more than the plane’s exact spot.

The altitude trace has its own grammar, which is in how to read a flight’s path. What can’t be tracked at all is covered in what’s that plane overhead.

When you’re following a flight in Skycap, the tracker labels the difference: “Live” when a position came in, “Now” when the plane is estimated from the schedule. Live position covers US airspace, and beyond it you get schedule and weather. How it works

How to read a flight's path: altitude, speed and the step climb

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

An altitude graph has four parts: a steep climb, a long flat cruise, a slow descent and a flat stretch or two near the end. Anything else on it has a reason. Most of the time the reason is fuel, wind or traffic, and most of the things that look odd are routine.

The staircase in cruise

A long-haul graph often climbs in steps: 31,000 feet for two hours, then 33,000, then 35,000. That’s a step climb, and it’s the plane getting lighter.

A heavy plane can’t reach its best altitude right after take-off. As it burns fuel, the best altitude rises, and the most efficient flight would be a very slow climb the whole way. Air traffic control works in levels, so airlines climb in steps as they get permission.

Those steps tend to be 2,000 or 4,000 feet because of the direction rule. Levels are stacked 1,000 feet apart, with eastbound flights on odd thousands and westbound on even, as the usual pattern above 29,000 feet. So a plane heading east from 35,000 goes to 37,000, not 36,000.

Reading a descent

The start of the descent is called top of descent, and on a graph it looks like a gentle corner. A jet comes down far earlier than most passengers expect. The rough pilot’s rule is 3 miles forward for every 1,000 feet down, so from 35,000 feet that’s about 105 miles out.

The final approach is steeper in appearance than it is. The usual glide path is 3 degrees, about 320 feet of height for each nautical mile.

Flat steps in the descent are normal. They’re usually a controller holding the plane at a level to keep it clear of other traffic, and they get more common near busy airports. A tiny climb in the descent is unusual but not alarming. A real climb on final approach is a go-around: the crew has decided to land on the next try, and it’s a maneuver they train for.

Spotting a hold

On the map, a hold is a racetrack: two straight legs joined by two turns, usually repeated. The standard pattern turns right, with an inbound leg of one minute at or below 14,000 feet and a minute and a half above that. On the altitude graph it’s a flat line where a descent should be, sometimes stepping down a level at a time.

Holds mean the airport can’t take you yet, often because of weather or a runway change. A tracker’s map often shows it before the airline’s estimate moves.

Long S-turns or a wide loop on the way in aren’t holds. Controllers lengthen the path to space planes on approach, and it’s common at hubs.

What the speed line tells you

Trackers usually show ground speed, which is the plane’s speed over the Earth. That’s why the same jet can read far faster eastbound than westbound on the same day, at the same power. Its speed through the air has barely changed. Wind accounts for most of the difference.

A sudden jump in ground speed at cruise usually means a change in wind or a turn relative to it. Speed falls steadily through the descent. Near the airport, there’s a step down as the plane slows for flaps. Don’t read these as anything the crew did in a hurry.

Where the graph lies

The altitude in a broadcast is pressure altitude, set to a standard reference, not true height above the ground. The FAA’s flight levels are defined that way, because every plane at a level uses the same reference. That means a flight showing 35,000 feet isn’t at exactly 35,000 feet of real height. It also means a graph near a mountain airport says little about clearance from the ground.

Gaps and straight lines usually mean coverage, not behaviour. An assumed path has no information in it. If a graph looks flat for an hour over the ocean, it may be a guess.

A sudden spike to zero or to some absurd height is a bad data point. A plane doesn’t do that. Ride bumps don’t show up on these graphs either: turbulence moves a plane by far less than the scale of the chart, so a flat line tells you nothing about how the ride felt.

Terms used here are in the flight tracking word list. Once the line reaches the runway, the wait isn’t over, and this is what’s left. If you’re watching someone’s flight from the ground, why the tracker and airline can disagree will save you a phone call.

To see what a step in your own flight was, Play the flight in Skycap replays the whole trip with the route, the ride and the view. What it shows in the air

How to share a flight with the person picking you up

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

Send a link that updates by itself, and send it before you board. The person waiting needs to know when you land and when you’ll be outside, and they need it without installing anything or texting you at 30,000 feet. A flight number in a text works too. A photo of your boarding pass doesn’t, and it gives away more than you’d think.

The bad version of this is familiar. You text “landing at 6:48” from the gate, the flight leaves 40 minutes late, you lose signal, and someone spends an hour in a cell phone lot watching for a text.

What the person waiting needs

“Landed” isn’t what they need. They need when the plane should land, which moves, and when you’ll be at the curb, which depends on your row, your bag and your airport.

A flight status page gives them the first. Almost nothing gives them the second. So a link to the airline’s status page is better than nothing and still leaves them guessing. If they’re picking you up, send when to leave for the airport along with it, and tell them whether you checked a bag.

Your options, from least to most useful

Method	What they get	The catch
Text the flight number	Any search or tracker finds it	They check by hand, and may pick the wrong day’s flight
Airline’s own share feature	Status, sometimes an itinerary	Differs by airline, and may need their app
Share your location	Where your phone is	It stops when your phone goes offline, so it goes dark for the whole flight
A live flight link	When you land, updating as it moves	They must open it, and anyone with the link can see it

Location sharing is the one people reach for, and it’s the wrong tool for a plane. It’s perfect again the moment you’re off the aircraft and walking, so use it for the last ten minutes.

Don’t send the boarding pass

The barcode on a boarding pass is built to an IATA standard, and it’s readable by anyone with a scanner app. It holds your name, your flight and your seat. It holds your booking reference, and airlines can add other data to it, such as a frequent flyer number.

A booking reference and a surname are, on many airline sites, enough to open your booking. In 2015, security journalist Brian Krebs described a reader who used a friend’s posted boarding pass to get into their airline account, where he could see upcoming flights and could have changed them. The code on the pass can be a key to your trip.

The same goes for posting one online, and for leaving a paper one in the seat pocket. If you want to show off the window seat, crop the barcode out or don’t post it until the trip is over.

A flight number has none of that. Flights are public. Anyone can look one up, and sharing it gives away only that you’re on it.

A short privacy check

Before you send any link, ask what’s in it and who can reach it. Good answers: no seat, no booking code, nothing that opens your reservation, and a way to stop it.

Then remember what a link is. It works for whoever has it. If you post it publicly, the public has your arrival time. That’s fine for a parent, less fine on social media while your house is empty.

When it goes wrong

If the flight is delayed or diverted, a live link updates and a text doesn’t. If the flight is cancelled, tell them yourself, since your new flight won’t be on the old link. A link for the original flight says cancelled and stops there.

If you’re the one following, keep the link open and don’t leave for the airport off the scheduled time. A tracker and an airline can show different things at landing, and that’s normal. On long flights, your own phone’s Live Activity may stop before the follower’s link does.

For the person at the curb, Skycap makes a link they can open with no account. It shows when you should land, when you’ll be outside and which door, and it carries no seat or booking code. You can end the share at any time, and otherwise it stops a day after the scheduled arrival. How sharing works

What's that plane overhead? Its own radio will tell you

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

Open any flight tracking app with a map, find your own location, and look for the moving dot that matches the plane’s direction. Nearly every airliner above you is announcing its identity, position, altitude and speed by radio, about once a second. Your phone is reading a copy of that.

The system is called ADS-B. A plane works out where it is from satellites and broadcasts it, unprompted, to anyone listening. Air traffic control listens. So do ground receivers run by companies and hobbyists, who pass what they hear to the tracking services. In the US, aircraft have needed it since Jan 1 2020 in most of the airspace airliners use, including everything above 18,000 feet.

No camera recognises the plane. An app knows your location and which way your phone faces, then draws a label where the broadcast says the plane should be.

Finding the right dot

The dot straight above you is rarely the plane you’re looking at. A jet at 35,000 feet is 6.6 miles up, so most of what you see is out to the side and a long way off.

Height above the horizon	Distance to a jet at 35,000 feet
Straight up	6.6 miles
45 degrees	9.4 miles
30 degrees	13 miles
10 degrees	38 miles

Those are my own numbers, from the angle and a flat Earth, which is close enough at this range. A fist at arm’s length is about 10 degrees, so a plane one fist above the horizon may be over the next county.

To pick between two dots, use altitude. A contrail means high, usually above 30,000 feet. A plane with visible flaps and a growing shape is low, probably within a few miles of an airport. Then check the direction of travel on the map.

The sound won’t help. From 35,000 feet straight up, noise takes about half a minute to reach you, and in that time the plane has moved about 5 miles. It always sounds behind where it is.

If you’re pointing a phone’s camera at the sky, expect the label to drift a little. The compass in a phone is thrown off by a metal case, a car or a railing. Move away from them, turn slowly and give it a second.

A jet at cruise is smaller than it looks. A 737-800 at 35,000 feet spans about 0.2 degrees, less than half the width of the Moon.

What won’t be on the map

Some aircraft are overhead and invisible, and that’s normal.

Military flights are the main group. The rule lets government operators switch ADS-B off on sensitive missions, and plenty of military traffic doesn’t broadcast at all. If you hear a loud jet and can’t find it, that’s usually why.

Small aircraft flying low may not carry it. Gliders and balloons are exempt in much of the airspace, and a light plane below 10,000 feet away from busy airports doesn’t need it. Helicopters at low level can be patchy for a different reason: receivers need line of sight, and buildings and hills get in the way.

Some aircraft appear with the name held back. Owners can ask the FAA to limit what’s shared of their flights. They can also use a temporary address that isn’t linked to the registry, so a receiver sees a plane without seeing whose it is. Those show up as a blank or a private entry, and the position is still there.

And some planes show with no route. The broadcast gives a call sign and a position. It says nothing about origin or destination, so a tracker has to match the call sign to a schedule or a flight plan. A cargo flight or charter may have no match.

Coverage matters too. A map can be thin where few receivers sit: open ocean, deserts and some rural areas. The plane is still broadcasting, and nobody nearby is listening.

How flight trackers work covers how a broadcast becomes a dot, and why it lags. With two trails overhead and only one lingering, the air decides. If someone’s about to land, here’s when to leave to pick them up.

With a plane going over, open Overhead in Skycap and hold up your phone. Names sit beside the planes on the camera, or you can read a board of what’s above you, with each one’s route and altitude. What Overhead shows

Why some planes leave trails and others don't

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

The air usually decides. A trail forms when hot, wet exhaust hits air cold enough to freeze it, which at cruising height is about −40 °C or colder. Whether it lasts depends on how much water vapour the surrounding air already holds. In dry air it’s gone in seconds or minutes. In air humid enough for ice to grow, it can spread for hours.

Burning jet fuel makes water vapour. A contrail is a cloud made by mixing, the same way your breath shows on a cold morning. Your breath doesn’t stay long on most mornings because the air around it is dry enough to take the water back. Same idea, 7 miles up.

Three balloons on one evening

The National Weather Service launches weather balloons twice a day, and each one reports temperature and moisture on the way up. I took three from the launch of 00:00 UTC on Nov 7 2026, which is the evening of Nov 6 in the US, and worked out humidity with respect to ice near cruising height.

Launch site	Height	Temperature	Humidity over ice
Upton, Long Island	35,445 ft	−52.4 °C	134%
Oakland, California	35,402 ft	−47.3 °C	79%
Lincoln, Illinois	34,352 ft	−53.1 °C	62%

All three are cold enough for a trail to form. So a jet in any of them could leave one. What differs is the last column. Over 100% means the air holds more vapour than ice can tolerate, so ice crystals in the trail grow instead of shrinking.

Over Long Island, a trail that evening should have lasted and spread. Over Illinois, it should have faded fast. Oakland sits between, and it shows something more useful.

Why two planes in one sky differ

The same Oakland balloon passed through a thin wet layer. At 33,392 feet the humidity over ice was 111%. At 33,554 feet, 162 feet higher, it was 89%. Ice-saturated layers can be that thin, and they have edges in every direction.

That’s your two contrails. One plane cruises inside the layer, and its trail hangs on and widens. Another crosses the same sky 2,000 feet higher, in drier air, and leaves a short line that stops almost behind the tail. Engines differ a little, and a modern engine can make trails in slightly warmer air than an old one. But for two jets above you at the same moment, altitude is the first thing to check.

It also explains a trail that starts and stops. The plane hasn’t changed anything. It went in and out of a patch of humid air.

Other things you might be seeing

Short white streaks from the wingtips, usually on a damp day near an airport, are different. Low pressure over the wing cools the air enough to condense it for a moment. NASA scientist Patrick Minnis calls these aerodynamic contrails, and at low height they’re short-lived liquid water.

A trail that drifts sideways and widens into a sheet is a persistent one sheared by wind at height. It can turn into something that looks like natural cirrus. Researchers can’t always tell them apart from the ground either.

And a plane with no trail at all may be flying well below 30,000 feet, where the air is usually too warm. If you’re checking, a tracker’s altitude number tells you.

How I got the numbers, and how far to trust them

Humidity over ice comes from each balloon’s temperature and dew point, using a standard vapour-pressure formula over water and another over ice. The soundings came from a public archive.

Balloon humidity sensors are at their worst in this cold. Treat 134% as “well over 100”, not as a measurement to the percent. Some of that Long Island layer may have been real cloud, which means you might not have seen a trail in it at all. A balloon also drifts, and it samples one thin column of air once. And I picked these three launches to make a point, so they say nothing about climate.

Whether a trail forms in marginal air depends on details of the engine and the exact temperature, and forecasts of these layers are still poor. The FAA and NASA both study them, because lingering trails trap some heat.

What’s that plane overhead shows how to find a plane’s altitude, and how to read a flight’s path reads the line once you’ve found it. If you’re at the airport waiting, this is the next hour.

With two trails above you, Overhead in Skycap names each plane and gives its altitude, so you can see which is higher. The trail on its camera view is a drawing, shown above 26,000 feet by day, and it doesn’t check whether a real one is forming. What Overhead shows

Why the tracker and the airline disagree

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

They’re usually timing different things. A tracker says “landed” when the wheels touch. An airline’s “arrived” means the plane reached the gate. The gap is taxi-in, 7 minutes at the median US airport and 18 at O’Hare, and it can be more than 50 at O’Hare on a bad evening.

Everyone in aviation splits a flight into four moments, and the arguments mostly come from people quoting different ones.

Moment	What happened	Who knows first
Out	The plane left the gate	Airline
Off	The wheels left the runway	Anyone watching it climb
On	The wheels touched the runway	Anyone watching it land
In	The plane parked at the gate	Airline

US airlines report these to the government. Under 14 CFR 234.4, a flight’s arrival and departure times are the times it reached and left the gate. A gate arrival counts when the pilot sets the parking brake at the gate, with the door opening as the fallback. Wheels-off and wheels-on are reported as separate items.

So the schedule you booked is a gate schedule. The FAA’s glossary agrees for scheduled airlines: an ETA is the time the flight is estimated to reach the gate. A plane that touches down at 7:10 for a 7:05 arrival is on time if it parks before 7:20. One that touches down at 7:05 at O’Hare can park at 7:30 and count as late. “Late” means gate time, 15 minutes or more after the published arrival.

The common mismatches

If the tracker says landed and the airline says in flight, the plane is probably on the ground and taxiing. An airline’s app may be waiting for the gate event, or may refresh more slowly. Give it several minutes at a big airport.

Sometimes a tracker says landed and the plane hasn’t. Some apps call a flight landed when its signal disappears near the airport, or when the scheduled time passes. Low planes drop out of receiver range all the time. I’ve made this mistake myself: if you only watch for a position vanishing, you’ll land a plane that’s still on approach.

An airline can also say arrived while the plane is still moving, when an estimated gate time sits on screen as if it were a fact. The reverse happens too. A gate event can be late, wrong or missing, so the arrival may post after you’ve left the plane.

“Departed” on an airline’s screen usually means pushed back from the gate. The plane can be in line for the runway for a long while after that, and a tracker that times take-off will show a later start.

The flight numbers may not match either. Trackers often show the operating airline’s call sign, and airline sites show the number you bought. A flight sold by Delta can appear under the regional carrier flying it.

Which to believe for what

For “is it in the air” and “has it touched down”, believe a tracker that has seen the plane. Position reports come from the aircraft itself.

For the gate, the belt, the boarding time and anything about your ticket, believe the airline. The airline owns those facts, and a tracker has them second-hand. Before you change a plan, check the airline’s app. That goes double for a cancellation or rebooking.

For “when will they be outside”, neither one answers it. Landing starts the clock, and what’s left depends on the row, the bag and the walk. A long stretch between on and in is explained in why O’Hare takes 18 minutes.

“Should have landed” is the honest phrase

When nobody has seen the plane land, the only honest statement is an estimate. I’d rather an app said “should have landed about 6:48” than “landed”, and I’ve tried to hold to that. A clock passing a scheduled time isn’t evidence of anything. On a diverted flight, it’s actively wrong.

If you’ve got a dot that stopped moving short of the runway, or a flight that says landed when the person hasn’t called, here’s how to read what a tracker is telling you. Abbreviations are in the flight tracking word list.

When you’re waiting on a flight, Skycap says “should have landed” until something has seen it land, then gives the time. It keeps the wheels-on time and the gate time apart, and every estimate reads as one. How it works

Every noise a plane makes, in order

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

Nearly every sound on a normal flight is a machine doing a scheduled job, and they come in the same order each time. The loudest thump of the climb is the landing gear going away, seconds after liftoff. The sudden quiet a little later is the engines being turned down on purpose, and the plane is still climbing.

Times below are rough estimates for a 737 or A320 on a typical flight, based on published climb and descent rates. Airline, aircraft, airport and weather all shift them.

At the gate

Thuds under your feet

Bags going into the hold, and then the hold door closing.

The air stops or changes

On most jets the air that starts the engines is the same air that feeds the cabin vents, so the airflow can drop or go quiet for a minute or so around engine start. It comes back.

Pushback and taxi

A clunk and a small jolt

The tug connecting and starting to push. Then a rising whine, one engine at a time.

A whirring from the wings

The flaps moving to their takeoff setting, often during taxi. You can watch them slide out from a window behind the wing.

A dog barking under the floor

This one is real, and you’ll mostly hear it on Airbus single-aisle jets. Passengers call it the barking dog. The usual source is the power transfer unit, a device that lets one hydraulic system pressurize another, and it can run when a crew taxis on one engine to save fuel. It comes in short bursts near the wings, and you may hear it again after landing. It’s routine.

Regular thumps

The nose wheel crossing joints and lights in the pavement.

Takeoff

A rising roar, then a press into your seat

Takeoff thrust. The thumping speeds up and the whole cabin rumbles.

The rumble stops

That’s the wheels leaving the ground. It often feels as though the plane has gone strangely smooth, because it has.

A heavy thunk and a rattle, seconds after liftoff

The landing gear comes up once the plane is climbing, and the doors close over it. This is the loudest mechanical noise of the climb.

The first few minutes

The engines go quiet, and you feel as if you’re sinking

The most startling moment of the climb for many people, and the most routine. Crews cut power after takeoff to reduce noise over the neighborhood and save the engines. The FAA’s noise-abatement profile sets that no lower than 800 feet above the airport, and the height varies by airline and airport. The nose lowers a little with it, and your stomach reports a fall. The plane is still climbing.

Whirring again, plus a small settling feeling

The flaps coming back in, usually in stages, over the first few minutes as the plane speeds up.

A ding or two

Most chimes are crew signals: a call between the flight deck and the cabin, or a change in the belt sign. Each airline has its own code.

A push of speed, about 4 minutes in

Passing 10,000 feet, where the US 250-knot limit ends, the plane accelerates and the nose lowers again. This is also where the rule against unnecessary talk on the flight deck eases off, which is why you’ll often hear from the crew about now.

Cruise, from about 18 minutes

The engines settle, and what’s left is a steady rush of air and ventilation. It sits there for the rest of the flight, loud enough that you stop noticing it.

A change in engine note, mid-flight

A long flight will often climb a little as it burns fuel and gets lighter, or change altitude for a better ride or wind. The engines rise for a minute or two and settle again.

The way down

The engines go quiet again, roughly 25 minutes before landing

The start of the descent. Jets come down with the engines near idle, so this is the plane gliding on purpose. Your ears may notice next.

A rumble and a shiver from the wings

Speed brakes: panels that rise on top of the wing to add drag. The FAA’s handbook says there’s usually a certain amount of noise and buffeting. If you’re at a wing seat you can watch them lift.

Whirring in stages

Flaps going out again, for the slower speeds of the approach. Each stage can feel like a gentle pitch change.

A thud and a much louder rush of wind, a few minutes before touchdown

The landing gear going down. The wind noise stays until you land, because the gear is now hanging out in it.

Landing and after

A bump, then a roar

The touchdown, followed by reverse thrust. The engines are louder after landing than they were on the approach, and this is deliberate: the handbook explains that the exhaust is redirected to help slow the plane, most effectively at high speed. The panels on the wing pop up at the same moment to put the plane’s weight on its wheels so the brakes work.

The roar stops and the brakes take over

You’ll feel it as the seatbelt pressing into you.

The barking again, on an Airbus

Then the engines wind down at the gate.

If a sound isn’t here, look at the flight attendants. They hear this every day, and they’re the best instrument in the cabin. For bumps, there’s why the first twenty minutes are bumpy. If sounds are a large part of why you dread flying, this is what the studies say helps. And for something else to do on the climb, work out which side the sun will set on.

The sounds are the same on every flight, and the bumps aren’t. At the gate, Skycap lists the first twenty minutes of your flight, with what’s forecast on the climb. See how the flight page works

Fear of flying: what the studies say helps

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

The approach with the best evidence is exposure: practicing being on a plane, in steps, with someone who knows how to run it, usually alongside training in managing anxiety. In a 2006 trial, 76% of people who had that treatment flew on a test flight afterward, against 20% of people on a waiting list.

Information helps too, by less than the internet suggests. Most tips lists open with how safe flying is. If you’re afraid, you’ve probably read that already. A fear is rarely settled by arithmetic.

What the trials did

The 2006 trial, led by Barbara Rothbaum at Emory, randomly assigned 83 people. Everyone treated had four sessions of anxiety management, then either time in a virtual airplane or time on a real parked airplane at the airport, over six weeks. Both treated groups did equally well, and at 6 and 12 months more than 70% of those who responded were still flying. An earlier trial by the same group, with 49 people, found 93% of treated participants had flown within six months.

A German program went further. In a 2020 study, 138 people with clinically significant fear had one group session of education and a breathing technique, then took a 2½-hour flight together on a chartered A320. Six months later, 55% were rated fully remitted and another 16% partly. There was no comparison group, so read that as promising.

The format seems to matter less than the practice. One trial of 60 people found virtual reality, computer-aided exposure with a therapist, and self-guided computer-aided exposure all helped, with no clear difference at a year. A small 2025 trial of 26 people found plain video exposure held up against virtual reality two years on.

What explanation does

Knowing what a bump is does count for something. A 2025 master’s thesis at Mississippi State showed students a short video about atmospheric turbulence before a study-abroad flight. Their general and anticipatory anxiety fell, mainly among those who weren’t phobic. Phobic students gained less, though few of them took part, and anxiety rose a little after the flight.

It’s one small study, and the direction fits an older German trial of 45 patients. There, people given cognitive treatment alone didn’t improve, while those who also had exposure in virtual reality did, both right after and at six months.

I read those together as: explanation can take the edge off ordinary nerves, and it helps you make sense of the flight, but a real fear usually needs practice as well. That’s my reading, and two studies don’t settle it.

Pages like every noise a plane makes and why the climb gets bumpy do that first job. Turbulence and unknown sounds were among the top causes of flight anxiety in a 2015 Norwegian survey, so they’re worth explaining. Treat those pages as background, and skip any that make you more wound up. If you’d rather give your eyes a job, what you’re flying over is a good one.

What it doesn’t cover

Most of these studies are small. People who sign up for a fear-of-flying trial want to fly, which may not describe you. Some of the programs are commercial, and I can’t vouch for any particular one.

I also haven’t covered medication or alcohol, which many people try. That’s a conversation for you and your doctor, and it’s outside what this page can tell you.

Forecasts are a separate question, covered in how to read a turbulence forecast. People who treat flight anxiety warn that checking one again and again feeds the worry.

When to talk to someone

If you’re cancelling trips, turning down work or missing family because of flying, that’s worth raising with a clinician. You wouldn’t be unusual. In a study of Norwegian passengers, about 1 in 22 had cancelled a flight because of anxiety in the previous two years, and those are people who were on a plane the day they were asked.

A psychologist who works with phobias can tell you whether exposure-based treatment suits you, and where to find it. Nothing here is a diagnosis or a treatment plan.

In surveys of the Norwegian public, likely flight phobia fell from 8% in 1986 to 3% in 2015. The surveys can’t say why, but the number moved.

How to read a turbulence forecast without scaring yourself

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

Read a turbulence forecast once, the night before, for one thing: does this flight look calmer, typical or bumpier than most? A forecast can say that with some skill. It can’t say what your seat will feel at 7:42, and any map that looks as though it can is promising more than the science has.

What the forecast is

The FAA’s handbook describes the main US public forecast plainly. It runs more than 10 turbulence calculations on weather model data, checks each against recent reports from aircraft, and weights them by how well each is matching. It’s issued hourly, hour by hour for three hours and then every three hours out to 18 hours, with no forecaster editing the result.

That description explains most of what the forecast can’t do.

It stops at 18 hours. Anything you look at days ahead is coarser than this, or a guess.

It gets better with more aircraft reporting. The handbook says it improves in daylight and where traffic is heavy, so a red-eye over the ocean is its weakest case.

It doesn’t specifically forecast bumps from convective clouds or small-scale terrain. Thunderstorms get their own forecasts, and in the NTSB’s count convective turbulence was present in 64 of 111 airline accidents. A calm turbulence map tells you nothing about a shower on the climb.

What the numbers and words mean

The number is eddy dissipation rate. It measures how stirred up the air is, and it’s the international standard for automated turbulence reports. Because it describes the air and not the aircraft, the same value feels different in a regional jet and a 777. The public forecast is scaled to three aircraft weight classes for that reason. A consumer site quoting one number for “the route” is skipping that step.

The words are older, and they were written for pilots reporting to each other. They’re descriptions of what happens in the cabin:

Pilot’s word	What the FAA says happens inside
“Light”	Slight strain against belts. Food service can continue, and walking is little or no trouble.
“Moderate”	Definite strain against belts. Loose objects move. Walking and food service are difficult.
“Severe”	Thrown hard against belts. Walking and food service are impossible.

Notice what the middle row says. The handbook adds that in “moderate” the aircraft stays in positive control at all times. A word that sounds alarming on a map means the coffee stops.

How good it is, tested

I checked forecast skill on the kind of forecast a flyer would actually see. Forecasts issued 3 hours before departure on 17 busy US routes were compared with 3,405 pilot reports from about a day and a half in early November 2026, all at 20,000 feet or higher.

Given one report of real bumps and one report of smooth air, the forecast ranked the bumpy one higher about three times in four. (The statistic is an AUC of 0.75, where 0.5 is a coin flip.) That’s useful and well short of certain.

The test also showed how a label that looks exact can mislead. A rule that called a flight rough whenever any single point on the route crossed a line fired on 64% of flights, and on 72% of flights where pilots reported only smooth air. So I dropped it. Rating the whole flight against a typical one held up better: on the same 572 flights, 53% of those pilots found bumpy were rated bumpier and 14% calmer, while 26% of calm flights were rated bumpier.

Read that last figure the right way round. About one calm flight in four gets the word “bumpier”, and some bumpy flights get “calmer”. I also set that scale using these same flights, so a fresh test would likely come in lower, and a day and a half in November isn’t a year.

That’s why I think “calmer, typical, bumpier” is the honest scale. It claims about as much as the forecast knows.

How to look at one

Two people who work with anxious flyers, Alex Gervash and Lawrie Jones, make the same case: checking again and again makes things worse. You get a moment of relief, and then the doubt comes back. A “smooth” forecast also sets you up for any bump to feel like a betrayal. Both have something to sell or a practice to run, and I think they’re right anyway.

So look once. Note the shape: whether the bumps are forecast early, late or somewhere over the mountains. Then close it. Rechecking at the gate adds very little, because the crew has newer information than you do, including pilot reports from the flights ahead. They can ask for another altitude. You can’t.

What a forecast can do for you is give the bumps a place in the flight. “About 20 minutes over the Rockies, then calm” is easier to sit through than a shake with no explanation. If you get a bump the forecast didn’t mention, the forecast was wrong, and nothing else has changed.

Whatever it says, the part you control is the same: the belt, the whole time you’re seated. If you’re flying a route with a reputation, the most-turbulent lists say less than they seem to. And for a stretch of forecast bumps at night, the window beats the seatback map.

Skycap gives you one outlook the evening before, in words, and says so if it changes. See how the ride forecast works

"Is turbulence dangerous?" What the NTSB numbers say

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

With your seatbelt on, turbulence very rarely hurts you. The NTSB went through every turbulence accident on US airlines from 2009 through 2018 and counted 123 people seriously injured. One of them was documented as wearing a seatbelt.

People do get hurt in turbulence, and pretending otherwise doesn’t help anyone. Most of them work on the plane, so the useful numbers are who got hurt and what they were doing at the time.

Who the 123 were

	Seriously injured	Belted	Not belted	Not recorded
Flight attendants	97	0	81	16
Passengers	26	1	17	8

Flight attendants are nearly four in five, because they are on their feet with a cart while you are sitting down. The NTSB’s count of where they were points the same way: 56 of the 97 were hurt in an aft galley.

Passengers were more often on the way somewhere. The largest group whose activity is known, 7 of the 26, were using, waiting for, or walking to or from a lavatory. For 13 of the 26, the record doesn’t say.

Two cautions on that table. There are 24 people with no belt record, so “one” is the documented figure and the true figure could be higher. The report also describes a passenger on a 2019 flight who was seriously hurt while wearing a belt with an extender. The NTSB read all of this and concluded that a seatbelt reduces the risk of serious injury for everyone on board. Its word is “reduces”. It’s the right word.

“Serious” has a set meaning here: more than 48 hours in hospital, or a broken bone other than a simple fracture of a finger, toe or nose, or a short list of other injuries. A bruise doesn’t count, so more people were hurt than 123.

What happened to the planes

In those 111 accidents, not one plane had substantial damage. In 106 there was none at all. The other five had minor damage, mainly to cabin parts such as seats, ceiling panels and overhead bins.

An “accident” in NTSB terms means someone was seriously hurt or the plane was substantially damaged. Every one of the 111 qualified on the first test. That’s why turbulence is the most common kind of US airline accident, at 111 of 295, a number that reads worse than what it measures.

Deaths

The NTSB report, published in 2021, lists the most recent death in a turbulence accident on a US airline as December 28, 1997: one passenger on a 747 flying from Narita to Honolulu. I searched for a later one and didn’t find it, but a search isn’t a count, and the FAA’s yearly table lists injuries, not deaths.

The case most people remember wasn’t on a US airline. On May 21, 2024, Singapore Airlines flight 321 was at 37,000 feet over Myanmar when vertical acceleration went from +1.35 G to -1.5 G in 0.6 seconds. Investigators said anyone unbelted likely left their seat. One passenger died. Press reports of the final report, published in May 2026, give 56 serious and 23 minor injuries among the 229 on board.

The first report also records that the crew called the belt sign on eight seconds before that drop. This is why crews ask you to stay buckled whenever you’re seated, with the sign on or off.

Whether it’s getting worse

Somewhat, and from a small base. The FAA’s table, built from NTSB reports, shows 207 serious turbulence injuries on US airlines from 2009 through 2024: 166 crew, 40 passengers, and one not classed as either. Years range from 5 to 23, and the two highest are the last two, 2023 and 2024, at 20 and 23. Passengers in those two years numbered three each. Counts this small swing around. From 1989 through 2018, the NTSB found no obvious trend in the accident rate per flight hour.

The atmosphere is a different matter. A 2023 study of weather records found that “severe-or-greater” clear-air turbulence over the North Atlantic rose by 55% between 1979 and 2020, from 17.7 to 27.4 hours a year at an average point. That is a real increase. It’s also about a third of one percent of the hours in a year, and it comes from a model of the air, not from flights, which steer around what they can.

What I’d do

The common advice is that turbulence is harmless. It’s better described as harmless to the person who is sitting down and buckled, which you can arrange.

Keep the belt on, low and snug, the whole time you’re in your seat. Go to the lavatory when the sign is off and don’t linger in the aisle. If you’re traveling with a child under two, the FAA says the safest place is an approved child seat or device, and the NTSB asked for more work to get them used.

The reasons for the ride itself are in how to read a turbulence forecast and where to sit for the smoothest ride. And if you’re at a window with nothing to do but think about this, try reading the ground from 35,000 feet instead.

What these numbers can’t tell you

They cover US airlines, and the belt table stops at 2018. They count serious injuries only. They can’t give you the odds for your flight, and I wouldn’t trust anyone who says they can.

The night before you fly, Skycap says how the ride should feel compared with a typical flight, and where any bumps are forecast along your route. See how the ride forecast works

The most turbulent US routes, and why the list misleads

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

Denver to Jackson Hole usually sits at the top of the list, level with Albuquerque to Denver. In the version the Denver Gazette reported in August 2026, four of the five bumpiest routes in North America touch Denver. The list is a fair map of where the mountains are. It tells you very little about your flight on a given day.

Rank	Route	Average turbulence score	Distance
1	Denver to Jackson Hole	18.18	406 mi
2	Albuquerque to Denver	18.18	350 mi
3	Jackson Hole to Salt Lake City	17.68	205 mi
4	Denver to Salt Lake City	17.54	390 mi
5	Bozeman to Denver	17.22	524 mi

The scores are as printed there. I couldn’t open the original ranking, and other outlets have reprinted versions with the top two the other way round. Distances are my great-circle figures.

Why mountains top it

Wind crossing a mountain range sets off waves in the air, and they can reach far above the peaks. The FAA’s handbook says the effects can at times extend past 60,000 feet, so cruising at 35,000 doesn’t put you over the top of them. The waves grow with the height of the range and with the wind blowing across it.

A gentle wave is a slow rise and fall, with little to feel beyond small changes in altitude and speed. The bumps come when a wave grows enough to break, like surf. Denver sits right where the air comes down off the Front Range, and Jackson, Salt Lake City and Bozeman are all in the same mountains. In summer those same places add afternoon storms, as the air over high, dry ground heats up.

Why short routes top it

The other half of the ranking is arithmetic. The score is an average, and short flights have less smooth air to average against.

Public flight records put Denver to Jackson Hole at about 66 minutes in the air. A 20-minute rough patch is about 30% of that. The same patch on a New York to San Francisco flight, at close to 6 hours in the air, is about 6%. You’d feel the same 20 minutes on both, and only one of them makes the list.

Short flights also spend much of their time climbing and descending through the lower air, where the ground stirs things up. A long flight has hours in smoother air to bring its number down.

What the list can’t tell you

An average over a year hides the days. Those routes have plenty of calm mornings and some rough afternoons, and the list can’t say which you’ll get. Wind across the mountains decides it, and that changes by the day.

The ranking also looks more meaningful than it is. First and fifth place are 5% apart, so a small change in a year’s weather can reorder the top of the table. I wouldn’t change a booking over it.

It isn’t an injury list either. In the NTSB’s study of 111 US airline turbulence accidents, thunderstorm-type turbulence was present in 64 and clear-air turbulence in 32, with mountain waves a smaller share. A route’s reputation and its record are different things.

I can’t check these numbers against anything better.

If you’re flying one of these

Take the morning flight in summer, when the afternoon storms haven’t built. In winter, time of day matters less and wind is what counts. Sit near the wing, and keep your belt on. If you want to know how a forecast for one of these flights should be read, start here.

There’s also a real reward. Coming into Jackson from Denver, the great circle puts the Tetons on the right, about 10 miles off. The approach on the day can change which side they’re on, but the view is the reason people fly this route, and it’s worth a few bumps.

A yearly average can’t speak for tomorrow. Skycap looks at the forecast for your flight and says whether it should be calmer, typical or bumpier than most. See how the ride forecast works

Where to sit for the smoothest ride: over the wing

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

Sit over the wing, and stay out of the last rows. An airplane pitches around its center of gravity, which sits near the wing, so the seats there move least. The back of the cabin moves most.

The difference is real, and it’s smaller than what your seatbelt does. If you only do one thing, buckle it.

Why the middle moves least

A seesaw works the same way: the person at the pivot barely moves, and the people on the ends go up and down. When a gust hits, an airplane rises or drops as a whole, and it also rotates nose-up or nose-down around its center of gravity. In the middle you get the first motion. At either end you get both added together.

The back gets something extra. NASA’s simulation work, quoted by the NTSB, found that for short, sharp gusts aft-cabin accelerations were significantly higher, partly from pitching and partly because the tail itself flexes.

Rolling side to side barely separates seats on a single-aisle plane, since nobody sits far from the centerline. Window or aisle won’t change the ride in any way you’d notice. That part is my reasoning from the geometry, and I haven’t found a measurement of it.

What the measurements show

There isn’t much. The best number comes from a NASA flight test in a Boeing 757 with accelerometers at the pilot’s station, the center of gravity and the aft galley. A researcher told the NTSB that load factors ran 12% higher at the pilot’s station and 40% higher at the aft galley than at the center of gravity.

Keep that at arm’s length. It was never published, it’s one airplane type, and the NTSB says it may not generalize. But it agrees with what the people who work back there say. Pilots and flight attendants told the NTSB that the effects are usually worst in the aft cabin, and they called it common knowledge.

Injury records lean the same way:

Where flight attendants were, unbelted	Seriously injured
Forward cabin	6 of 45 (13.3%)
Middle cabin	2 of 42 (4.8%)
Aft cabin	67 of 129 (51.9%)

Those are 2009 to 2018 US airline accidents, so every one of these flights was already a bad day, and these are people on their feet. It says nothing about the odds on a normal flight. The NTSB also warns that the raw counts partly reflect where people spend time unbuckled, and the aft galley is where the work is. The gap is still wide.

How to pick the row

Open the seat map for your flight and find the wing, which many seat maps mark. On a 737 or A320 you’re looking for the rows around the overwing exits. Row numbers differ by airline and aircraft, so check yours.

The aircraft matters more than the row if you get to choose. A heavier airplane is pushed around less by the same air, which is why the public turbulence forecast is scaled to three aircraft weight classes. A 777 will usually ride better than a regional jet on the same day.

There are two costs. Sit over the wing and your view of the ground is partly wing, which is a real loss if you came for the sunset. The front of the wing gives you more of the view and keeps most of the ride.

The other is minutes getting off, and here the middle of the plane has nothing to apologize for. My deplaning model for a 737-800 or A320 at 82% full, counted from door open, gives 1.7 to 4.4 minutes from row 5, 4.6 to 7.9 from row 15 and 7.5 to 11.4 from row 25. That’s a model built from one person’s timed flights, and the method is here. The back of the plane loses on both counts.

How much it matters

For someone who dislikes bumps, I’d take a wing seat over an extra inch of legroom. For someone who is frightened of them, it helps a little, and understanding what the bumps are or what actually helps with fear will do more. A wing seat in a rough patch still feels like a rough patch.

If the ride is why you’re choosing this seat, Skycap shows how your flight is forecast to feel, stage by stage, before you board. See how the ride forecast works

Why the first twenty minutes are bumpy, and why it stops

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

The bumps after takeoff come from the lowest layer of the atmosphere, where the sun heats the ground unevenly and the wind trips over whatever is in its way. A jet is through that layer in minutes. The FAA’s handbook says the air above the tops of puffy clouds is generally smooth, and a 737 reaches cruise altitude about 18 minutes after takeoff.

So the rough part has an end you can often see out the window.

The ground stirs the air

On a sunny day the ground warms the air sitting on it, and it does so in patches. Bare soil and plowed fields get hotter than water or leafy ground, so bubbles of warm air rise from some places and cooler air sinks between them. Flying through that feels like driving on a road with potholes you can’t see. The handbook says these currents are most active on warm summer afternoons when winds are light.

Wind adds a second source. Buildings and hills break a smooth flow into eddies, which get stronger with wind speed and rougher ground. A windy day at a coastal or mountain airport can be lively in January, with no sun involved.

Early mornings and nights can have their own version, when a cold layer sits near the ground under faster wind. The bump there comes from the two layers sliding past each other, and it’s usually confined to the lowest few thousand feet.

Why it stops

Rising air keeps rising until it has cooled to the temperature of the air around it. If it cools enough on the way, it makes a cloud. So those flat-bottomed cumulus clouds on a summer afternoon are markers: the handbook calls them signposts of this kind of turbulence, and says the top usually marks about where the rising currents end.

That gives you something to watch for. Expect bumps below and inside those clouds, and expect them to ease as you come out the top. How high that is depends on the day and the place. On a dry day there may be no clouds at all and the bumps still happen, with nothing to warn of them or show where they end.

The climb, by the clock

These are estimates for a Boeing 737-800, worked out from published climb rates. Real flights vary with weight and weather.

After takeoff	Altitude	What you might notice
2 minutes	5,000 ft	Usually the busiest part of the climb, with the plane turning and the flaps coming up
4 minutes	10,000 ft	The nose may lower a little as the plane speeds up
11 minutes	24,000 ft	Usually above the fair-weather clouds
18 to 19 minutes	35,000 ft	Level off, and the engine note settles

The nose lowers at 10,000 feet because below that height US rules cap speed at 250 knots. Above it, the jet accelerates. Some people read this as the plane leveling off or sinking. It’s still climbing. There’s more on that in every noise a plane makes.

If you get to choose

In summer, take the morning flight. It’s the one piece of schedule advice the physics backs: the ground hasn’t heated up yet, and the afternoon’s convection hasn’t started. In winter the time of day counts for less, and wind and fronts decide it.

When it doesn’t stop

Sometimes the bumps carry on past the clouds, usually for a different reason. High up, wind shear near the jet stream and waves off mountains can bump a plane in clear air, far above anything the ground is doing. Storm clouds also grow much higher than fair-weather ones, and crews steer around those. How to read a turbulence forecast covers those.

The same layers are waiting at the other end. About half of the turbulence injury accidents the NTSB studied happened during descent and approach, when flight attendants are often still up and working, though the NTSB notes that its figures don’t account for how long each phase lasts. The belt sign tends to come on for the way down. I’d have the belt fastened before that.

If bumps on the climb are what bother you most, it’s also worth reading where to sit. And for something better to look at once the wing stops shaking, here’s how to read the ground from cruise.

At the gate, Skycap lists the first twenty minutes of your flight: what’s forecast on the climb, and the altitude where the air should settle. See how the flight page works

Night flights: what you can see, including the northern lights

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

Yes, you can see the northern lights from a plane, and a plane has real advantages over the ground: you’re above most clouds and far from streetlights. You need a route that goes far enough north in magnetic terms, a dark sky, an active night, and a seat on the side facing the pole. Flying from North America to Europe overnight, that’s usually the left side.

The rest of a night flight is worth staying awake for too, if the shade is up and the cabin’s dark.

Your route’s magnetic latitude matters most

The aurora follows the magnetic pole, which sits over northern Canada, so distance from that pole counts for more than geographic latitude. NOAA’s rule of thumb puts the southern edge of the auroral oval at about 66° geomagnetic latitude on a quiet night, moving about 2° toward the equator for each point on the 0-to-9 Kp scale of geomagnetic activity. The oval can also be seen from hundreds of miles south of it.

I worked out the highest geomagnetic latitude on the great circle for some common routes, using the pole’s 2025 position:

Route	Peak geomagnetic latitude	Kp where the oval’s edge reaches it
New York to Paris	59°	About 3 to 4
New York to London	60°	About 3
Chicago to London	64°	About 1
Denver to London	68°	Inside the oval’s zone
New York to Reykjavik	69°	Inside the oval’s zone
Los Angeles to London	71°	Inside the oval’s zone
Seattle to London	77°	Often poleward of the oval

A surprise in that table is how much North America helps. London and Seattle are both at about 53° geomagnetic, even though London is about 4° farther north on a globe. At the same latitude, you’re magnetically much farther north over Canada than over Europe.

For comparison, New York to San Francisco peaks near 51°, so a domestic flight needs a big geomagnetic storm.

Which window

Over most of these routes the aurora is to your north, which is on your left going east and your right going west.

The high routes break that rule. Seattle to London reaches about 77° geomagnetic, well poleward of the oval on an ordinary night, so the aurora may be overhead or even to the south, on your right. If you’re on one of those routes, check both sides if you can.

Look north, low on the horizon. From 35,000 feet that horizon is about 230 miles away, which on a route near the edge can be the difference.

Darkness is the part people forget

NOAA says the best aurora tends to come within an hour or two of midnight, about 10 PM to 2 AM local time, and that activity peaks around the equinoxes. Summer at high latitudes rarely works, because it doesn’t get dark. In my numbers for a Seattle to London flight leaving in the evening on June 21, the sun barely reaches the horizon at any point.

Overnight eastbound flights suit this well. For a New York to London flight leaving at 9 PM on January 7, 2027, a date picked because it’s close to a new moon, I get peak geomagnetic latitude about three hours in, at around 2 AM local solar time, with the sun far below the horizon and no moon up.

The westbound daytime flights back from Europe mostly don’t work. A 4 PM departure from London in January stays in dusk-level twilight for much of the crossing, because the plane is keeping pace with the sunset.

Moonlight dims the aurora’s look. Full moons this winter fall around November 24, December 24, January 22 and February 20. New moons are around December 8, January 7 and February 6. Those dates come from the same calculation.

The Kp forecast is on NOAA’s Space Weather Prediction Center site. A high number is a reason to stay awake, with no promise attached.

What the aurora actually looks like from a seat

Dim, usually. A camera sees more green than your eyes will. Expect a pale band or arc at first, and let your eyes adjust for several minutes away from the seatback screen. Cabin reflections do the most harm: turn your phone over, lean close to the window, and hold a jacket or blanket around your head and the pane.

A clue that it’s aurora and not cloud: it may move slowly, and stars show through it.

Cities and what lies between them

City lights are the sure thing on a night flight. Western cities built on a grid look like a printed circuit, with straight bright lines running to the edge of town and then stopping. Older eastern cities look more like a spill. From altitude you can often see the interstates between towns as dotted lines of traffic.

Two sights are worth looking for because they don’t look like towns. North Dakota’s Bakken oil region shows up as scattered specks in otherwise dark country. NASA found that most of those specks are lights on drilling equipment and temporary housing, and a few are gas flares. Minneapolis to Seattle passes about 100 miles south of Williston on the great circle, so look right. Texas’s Permian Basin is the other, around Midland: a Houston to Los Angeles flight passes close on the right, and Dallas to El Paso close on the left.

In winter a full moon rides high around midnight, too high to see from a side window, and it lights cloud tops well enough to show their shapes. That’s a sight of its own, and I wouldn’t turn it down.

Where the numbers stop

The geomagnetic numbers use a simple tilted-dipole model of Earth’s field, and the oval changes shape through the night and during storms. Routes are great circles, and the North Atlantic tracks move every day, as why your flight path curves explains. I can’t predict clouds above you, a cabin that keeps its lights up, or the person in 14C who wants the shade down. For the ride on the same crossing, there’s how to read a turbulence forecast. For dusk on the way, the sunset article has the side.

For your flight, Skycap marks where the moon should be and which cities’ lights you should pass, window by window. It doesn’t forecast the aurora. See the window guide

What am I flying over? Reading the ground from 35,000 feet

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

At cruise you can see about 230 miles to the horizon in clear air, and the ground below is labeled if you know the shapes. Circles mean irrigated High Plains. A grid of roads a mile apart means you’re somewhere west of Pennsylvania.

Your phone can usually find its position in airplane mode by the window, so a map downloaded before you board will settle any argument. These shapes are more fun, and they work when the seatback screen is showing an ad.

A grid a mile on a side

Across most of the country from Ohio westward, public land was surveyed on a grid. Starting with Ohio’s Seven Ranges in 1785, federal surveyors laid out townships six miles square, each cut into 36 sections of about a square mile. Roads and field edges followed those lines, and they still do. From the air you get graph paper to the horizon.

It makes a ruler. Count the road crossings and you know how many miles you’ve traveled, or how fast. The system never reached the original 13 states, Kentucky, Tennessee or Texas, so the pattern changes over those.

The grid also has small kinks. Lines of longitude converge toward the pole, so a square grid can’t stay square, and surveyors built in correction lines where north-south roads jog sideways. Those are the little offsets you’ll see in straight county roads.

Circles

A green circle means a center pivot: a long pipe on wheels sweeping around a well. Most are half a mile across, because a quarter-mile pipe fits inside a 160-acre quarter section of that same survey grid. A 1970s NASA study of Finney County, Kansas, found 95% used that length, and some circles run to a mile. The corners of the square stay dry, which is why a High Plains field from above looks like coins in a tray.

They need groundwater, so they cluster above the Ogallala Aquifer, across western Nebraska and Kansas, eastern Colorado and the Texas Panhandle. If you’re seeing lots of circles, you’re probably over one of those. In summer each one is a different stage of crop, and some show as wedges where the pivot has stopped partway around.

Rivers tell you which way you’re going

The Lower Mississippi loops so hard that it has left its old loops behind as crescent lakes. The geologist Harold Fisk mapped those old channels for the Army Corps of Engineers, and his 1944 maps are worth finding if you like this kind of thing. From a plane you see the same thing live: a thick brown river, pale scars of former bends on either side, and a few oxbow lakes cut off from it.

Atlanta to Dallas is a good flight for this. The great circle crosses the river at about Greenville, Mississippi, a little past halfway, with Lake Chicot, the largest oxbow in the country, close by on the left.

North of about St. Louis the river looks different, with tree-lined bluffs and locks. It’s a good clue to your latitude.

Stripes in Pennsylvania

West of New York and Philadelphia the Appalachians show up as long, even stripes running northeast to southwest. NASA explains the color: forested ridges of tough rock, farmed valleys of softer limestone and shale. The rock was pushed from the side and buckled like a rug. The great circle from New York to Chicago runs close to State College, a little over a quarter of the way along. If you’re leaving Newark or JFK heading west, watch for these stripes in the first half hour.

Two colors of Great Salt Lake

Flights from the upper Midwest to San Francisco can pass right by Great Salt Lake. Look for a straight line across the water with a different color on each side. That’s a railroad causeway built in 1959. The rivers feed only the south arm, so the north arm got far saltier, and salt-loving algae and bacteria turn it pink, while the south stays blue-green. Since 1966 the north arm has averaged about 317 grams of salt per liter, against 142 in the south.

West of the lake, the white plain is the Bonneville Salt Flats. On a great circle from Minneapolis to San Francisco, both are within about a dozen miles on the right, about two-thirds of the way along.

Route by route

These are from great-circle calculations between the airports, and real tracks can run 50 miles or more to either side. Use the side as a good bet. Why your flight path curves explains the gap.

Flight	What to look for	Side	Roughly when
Seattle to Los Angeles	Mount Rainier, then Mount Hood	Left, then right	First 15% of the flight
San Francisco to Seattle	Mount Shasta, Mount Hood, Mount Rainier	Right	From about 40%
Los Angeles to Seattle	Yosemite, Tahoe, Hood	Left	From about 30%
New York to Los Angeles	Grand Canyon, almost underneath	Right	About 85%
Los Angeles to Chicago	Grand Canyon	Right	About 20%
Phoenix to Chicago	Circles near Garden City, Kansas	Right	About halfway
Chicago to Denver	The Platte River and circles in Nebraska	Left	About 65%
Dallas to Denver	Southwest Kansas circles	Right	About 65%
Atlanta to Dallas	The Mississippi and Lake Chicot	Left	A little past halfway
Los Angeles to Atlanta	Meteor Crater, Arizona	Left	About 20%

Clouds cancel all of this, and haze can cut that 230 miles to not much. On those days, an evening flight still has the sunset. At the height of a Plains summer afternoon, the puffy clouds that make the climb bumpy can sit right over the view too.

Skycap goes through your route before you board and tells you what you should pass and which window it’s out of. In the air, it shows you the sight nearest the plane. See the window guide

Which side of the plane for the sunset

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

Flying north, sit on the left. Flying south, sit on the right. Those two hold all year, because the sun sets in the west. Flying east, the sunset is behind you and no window gets a good view of it. Flying west, it’s in front of you, and the season decides which side has it.

For sunrise, swap them: right when you’re heading north, left when you’re heading south, and in front of you heading east.

The westbound case is where the usual advice falls down. “Left in winter, right in summer” assumes you’re flying due west, and you mostly aren’t.

New York to Los Angeles, month by month

A flight from JFK to LAX starts out heading about 274° and ends heading about 246°, because the shortest route bends toward the southwest as it goes. By the time the sun sets, the plane has usually turned to the left of due west, and that pushes the sun to the right of the nose.

I worked out where the sun sits at the moment it sets for an evening departure:

Date	Where the sun is at sunset
December 21	Ahead, slightly left
January 15	Ahead, slightly left
February 15	Straight ahead
March 20	Ahead, slightly right
April 15	Right
May 15	Right
June 21	Right, about 45° off the nose
August 15	Right
September 22	Right, closing on the nose
October 15	Straight ahead
November 7	Straight ahead

So on this route, book the right side from April to mid-September. From October to March the sun goes down close to dead ahead, and from a side window you get the color without the disk, so either side will do. In deep winter the left has a slim edge. The common rule gets that part right and misses the other seven months.

The same logic runs on other westbound flights. Chicago to Denver bends less, and my numbers put the sun about 40° right of the nose in mid-May and about 20° left in mid-January.

North and south are the easy ones

I ran the same sums across the year for three routes, and the side didn’t change once.

Route	Sunset side
Seattle to Los Angeles	Right, all year
Boston to Orlando	Right, all year
San Francisco to Seattle	Left, all year

If you only fly one direction of a route, this is the reliable kind of booking.

One worked example

Take a JFK to LAX flight leaving at 4 PM on November 7. On the ground at JFK the sun sets at about 4:46 PM. The plane keeps it for much longer. On my numbers it sets at about 7:17 PM New York time, 3 hours 17 minutes after takeoff, somewhere over southern Colorado, within a degree or two of straight ahead.

Two things stretch that sunset.

The first is height. At 35,000 feet the horizon sits about 3° below level, so at around 40° north the sun goes down 17 to 19 minutes later than it does on the ground beneath you.

The second is direction. Earth’s surface at 40° north turns east at about 795 mph, and a jet heading west at around 530 mph is running with the sunset, although it’s losing. By my calculation, on that November date the sun takes about an hour to go from 5° above the horizon to 6° below it for someone standing still at 40° north. Flying west it takes about three hours. Flying east, 37 minutes.

If you’d rather not have the sun at all

Around the middle of the day at mid-latitudes, the sun is to the south. So on a daytime flight across the US, it stays on the left going west and on the right going east. Every sample flight I ran, in November and in June, came out that way, apart from the last hour or so of an eastbound June flight, when the sun was behind. If you want to sleep or see a screen, take the opposite side: right going west, left going east.

Sunrise out of JFK to Europe works the same way as any eastbound flight: it’s ahead, a little right in January and to the left in June. If you’re staying awake for it, what you can see at night will fill the hours before.

How I worked this out

The sun’s position comes from standard solar position formulas, evaluated each minute along a great-circle route at 460 knots and 35,000 feet with no wind. Sunset is the moment the sun’s top edge drops below the horizon as seen from that height. I call anything within 20° of the nose “ahead”, because from a side window that’s a glow more than a sun.

There are more routes, month by month, on the sunset side page.

What the sums leave out

The real route won’t be the great circle, as why your flight path curves explains, and a headwind moves the time. It can’t see clouds. It also won’t know that your window seat is level with a wing, or has no window at all, so check the seat map too. If bumps matter more to you than the view, the smoothest seats are the ones over the wing.

Skycap works this out for the flight you’ve booked, from its route and departure time, and tells you which window to sit at and about when the sun goes down. See the window guide

Why your flight path curves on the map

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

Your flight path curves on the map because the map is flat and Earth isn’t. The shortest line between two cities is an arc around the globe, called a great circle, and most flat maps draw it as a bend. New York to Hong Kong by the straight line on a wall map is about 10,160 miles. By the great circle it’s about 8,060, and that route passes within 6° of latitude of the North Pole.

That’s the first bend. The flight you’re on has usually picked up a few more.

Take a string to a globe

Stretch a string tight between two cities on a globe and it shows the shortest route. For cities at similar latitudes in the north, it swings poleward, because that’s where the circles of latitude are small and the distance across is short. The flat map stretches those high latitudes so much that the shortcut looks like a detour.

The saving grows with distance and latitude. Here’s the comparison with the straight line on a standard Mercator map, using great-circle distances between the airports:

Route	Straight on the map	Great circle	Saved	Northernmost point
Chicago to Denver	887 mi	886 mi	1 mi	42°N
New York to Los Angeles	2,494 mi	2,470 mi	24 mi	41°N
New York to London	3,579 mi	3,443 mi	136 mi	54°N
Los Angeles to London	6,048 mi	5,444 mi	604 mi	62°N
Chicago to Tokyo Haneda	7,135 mi	6,292 mi	843 mi	63°N
New York to Hong Kong	10,159 mi	8,061 mi	2,098 mi	84°N

On a domestic flight the curve is real and doesn’t save much. Heading from New York, the great circle to Los Angeles starts at about 274° and ends at about 246°. That swing is why a sunset you’d expect on the left can turn up on the right.

Wind bends it next

Airlines don’t want the shortest route. They want the quickest and cheapest, and the air is moving. The jet stream blows west to east at speeds that can top 275 mph, so eastbound flights go looking for it and westbound flights try to stay out of it.

The North Atlantic is the clearest case. Traffic there runs in two tides, west by day and east by night, on a set of parallel tracks. Those tracks are redrawn every day to fit where the jet stream and pressure systems are. So your New York to London flight tonight and tomorrow night can take noticeably different lines across the middle of the ocean, and neither may be close to the great circle. On a strong jet stream day, eastbound flights can make a long swing to ride it. That’s a big part of why the same flight’s line looks different from one day to the next.

Then rules

A two-engine airliner can’t go anywhere it likes. US rules limit twin-engine jets to routes within 60 minutes of a suitable airport unless the airline is approved for longer. Those approvals, known as ETOPS, now stretch that limit by hours, and it still shapes long ocean routes. It’s one reason a Pacific crossing can stay closer to Alaska or Hawaii than the arc alone would.

Politics bends routes too. In March 2022 the US closed its airspace to Russian aircraft, and US airlines that used to cross Russia for Asia have flown around it since. On a map of the Arctic, you can see that detour.

Then everything else

Over the US, the arc gets chopped into pieces. Flights follow published departure and arrival paths near busy airports, and controllers space them out by giving them turns. Storms add more: a line of thunderstorms in the Midwest can put an entire stream of flights on a loop around it. Short flights can look oddly zigzagged because of this, and because the arrival path into a big airport may start well off to one side.

How to read the line in front of you

Three different lines get called “your route”, and it helps to know which you’re looking at:

The great circle is the geometry. Many maps draw this when they have nothing better.
The filed route is the plan the airline sent in.
The flown track is where the plane has actually been.

If the line ahead of the plane is a perfect smooth arc, it’s usually a guess. If it has corners, somebody filed it. And if your plane has wandered off the line, it’s most often weather or air traffic control doing their job.

That matters at the window too. A landmark list based on the great circle can be off by a whole side of the plane on a day the track runs 50 miles north. If the bends on your flight have you wondering about the bumps over the Rockies, that’s a wind story as well.

Where the numbers come from

Distances are my calculations from airport coordinates on a spherical Earth, so treat them as good to within a fraction of a percent. Real flights are longer than either figure.

If you’d like to see how your own route ran, Play the flight in Skycap replays it from takeoff to landing, with the ride and the view. See how it works

Why flying east is faster: about 40 minutes coast to coast, an hour in winter

Derek Johnson — Sat, 07 Nov 2026 00:00:00 GMT

Flying east is faster because the jet stream blows from west to east, and a plane’s speed through the air adds to the speed of the air itself. Coast to coast, that’s worth about 40 minutes over a year and about an hour in winter. The figure you usually read, “about 30 minutes”, is low for those flights, and it hides how much the season matters.

Those numbers are mine. I took 12 months of public flight records, September 2025 through August 2026, and compared the median time in the air each way. That’s wheels-off to wheels-on, so taxiing and schedule padding don’t count.

Route	Westbound	Eastbound	Difference	December to February	June to August
Boston and San Francisco	363 min	303 min	60 min	75 min	54 min
JFK and San Francisco	349 min	296 min	53 min	70 min	41 min
Atlanta and Seattle	293 min	248 min	45 min	69 min	33 min
JFK and Los Angeles	327 min	289 min	38 min	59 min	20 min
Chicago and Los Angeles	230 min	209 min	21 min	28 min	13 min
Chicago and Denver	128 min	114 min	14 min	19 min	11 min
Dallas and Los Angeles	165 min	154 min	11 min	20 min	-2 min

How it scales with distance

Across 373 US east-west routes between 1,000 and 3,000 miles, each with at least 600 flights a year in both directions, the median gap looks like this:

Route length	Routes	Over the year	December to February	June to August
1,000 to 1,500 mi	175	20 min	31 min	7 min
1,500 to 2,000 mi	94	32 min	47 min	15 min
2,000 to 2,300 mi	39	39 min	58 min	17 min
2,300 to 3,000 mi	65	41 min	63 min	28 min

So “about 30 minutes” is a fair figure for a mid-length route averaged over a year. On a long route in January it’s half the story. On a short one in July it’s four times too high.

What’s doing it

Jet streams are narrow bands of fast wind, usually about 30,000 feet up, right where airliners cruise. NOAA says they can top 275 mph and that they’re strongest in winter.

You can work out the wind a flight sees from these times. JFK to San Francisco is 2,580 miles on the great circle. At the median times, that’s about 444 mph going west and 523 mph going east, which means an average wind along the route of about 40 mph. In winter it comes to about 52 mph, and in summer about 31. Those are whole-flight averages, climb and descent included, and westbound flights choose tracks that avoid the worst of it, so the jet’s peak is far stronger than this.

Why the south loses it in summer

In summer the polar jet stream typically moves up toward the US-Canada border, and the table shows it happening. Boston to San Francisco keeps 54 minutes of its gap in summer. Atlanta to Los Angeles, on a route much farther south, goes from 45 minutes in winter to 5. Dallas to Los Angeles disappears altogether, and in those three months the eastbound flight ran 2 minutes slower.

For a single route the seasons look like this. JFK to Los Angeles has a median of 342 minutes in February and 308 in July. The trip back takes 279 minutes in February and 298 in July. The westbound flight gains 34 minutes from winter to summer, and the eastbound loses 19.

The Earth’s spin has nothing to do with it

A common explanation is that the planet turns beneath you, so flying east or west changes the distance. The air turns along with the planet, though, and a plane flies through the air. With no wind, both directions would take the same time.

North-south routes make the point, since they cross the jet stream instead of riding it. Minneapolis to Dallas has a median of 119 minutes one way and 117 the other. Seattle to Los Angeles is 130 and 133.

What it means for your day

Airlines build these differences into schedules, so the printed times already lean this way. What a schedule can’t know is this week’s wind. On JFK to Los Angeles, one westbound flight in ten took 307 minutes or less, and one in ten took 350 or more, across the year. That’s 43 minutes of spread on the same route, without a single delay on the ground counted.

If you’re meeting someone off a winter flight from the West Coast, it can arrive well before you’d guess. A westbound flight in January is the one to pad, especially if your connection is tight.

The line your flight takes depends on the same wind, which is covered in why your flight path curves. It’s also why a westbound sunset lasts so long: you’re running with it. And the edges of the jet stream are where much clear-air turbulence comes from, as how to read a turbulence forecast explains.

What the records leave out

The records cover US airlines’ domestic flights, so the Atlantic and Pacific aren’t here. Medians hide the unusual days, and air time also includes detours around storms and holds. This is one year, with one winter in it, and another year will differ. The route figures are medians for each direction, so a difference isn’t any single flight’s gap.

Skycap checks the winds along your route and says about how many minutes they should add or save on this flight. See what it shows in the air

Getting off a 737 takes about a minute and a half, plus 20 seconds a row

Sat, 07 Nov 2026 00:00:00 GMT

From the door opening, row 25 of a 737-800 usually steps off about 9.5 minutes later. The old rule of 30 seconds a row says 12.5.

A better rule for a full narrowbody is about a minute and a half, plus 20 seconds a row. That’s modelled, and tested against 30 timed flights.

The rule of thumb runs long at the back

Row	Early, 1 in 10	Typical	Late, 1 in 10	30 seconds a row
5	1.7	3.0	4.4	2.5
15	4.6	6.3	7.9	7.5
25	7.5	9.5	11.4	12.5

Those are minutes from door open, on a 737-800 about 82% full. The two lines cross near row 8. In front of that the old rule is a little short. Behind it, the rule runs long, by 3 minutes at row 25.

An A320 lands in the same place, at 3.0, 6.3 and 9.6 minutes for those rows.

Count rows from the front

What matters is the number of seats ahead of you, and ticket row numbers can skip. On one flight in the test set, row 27 on the ticket was the 15th row of the cabin.

The cabin layout moves it as well. At row 25, Southwest’s all-economy 737 runs about 10.1 minutes against 9.5 on a 737 with first class. JetBlue’s A320 runs about 10.3 against 9.6.

Your seat matters less than your row

At row 25, an aisle seat saves about a minute over a window. At row 5 it saves about half a minute. A middle seat is almost the same as a window.

Your own overhead bag comes out near zero, because you’re in a queue anyway. Treat that as a floor. The simulation puts every bag right at your row, and in practice some sit several rows behind.

A fuller flight costs more. At row 25, a flight 90% full adds about 0.9 minutes over 82%, and one 70% full saves about 1.4. Southwest flew about 75% full over the past year and Delta about 83%, so the same seat can differ by a minute from load alone.

A second open door would save about 5.7 minutes at the back of a narrowbody, but US flights rarely use one at a jet bridge.

How this was made

The model is a simulation of a cabin, one passenger at a time: who stands, who waits, who blocks the aisle. It runs on public timing studies and yearly load figures by airline. A short formula fitted to the simulation gives the numbers here.

I tested it on 30 real flights timed by one traveller. It missed by 0.88 minutes on average, against 1.13 for 30 seconds a row, and 83% of flights fell inside its early-to-late range.

Figure	Basis
The 30 timed flights	Measured, by one person
Minutes by row on narrowbodies	Modelled and tested
Seat, load and door effects	Modelled, not tested
The clock start at door open	Judgement, about a minute either way

What this can’t tell you

A straight line through those 30 flights does about as well as the simulation, so the test doesn’t prove the extra machinery is right. What the simulation adds is the effects of load and seat, and those rest on the simulation being sound.

One traveller timed all 30 flights, in 2016 to 2018, mostly from a middle seat. Nothing in the set is forward of row 7, so row 5 is an extrapolation.

Widebody and regional numbers aren’t tested at all, so they aren’t on this page.

American and United cabin maps aren’t available row by row, so on those airlines the row-to-seats step is rougher.

It doesn’t cover wheelchair assistance, gate-checked bags, stairs or buses at a remote stand. Behaviour since 2020 is unknown, and nothing here measures who stands up early.

Landing to the curb takes 16 to 36 minutes, depending on the airport

Sat, 07 Nov 2026 00:00:00 GMT

From touchdown to the curb, a carry-on passenger in row 25 of a 737 takes about 16 minutes at Burbank and about 36 at O’Hare. On a slow trip, 1 in 10, O’Hare runs to 55.

Those are modelled figures across 55 US airports, for a typical day. No airline, airport or tracker publishes this number, and the person circling the arrivals curb is the one who wants it.

Most of the slow trips start on the taxiway

Among the 12 busiest airports, the typical trip runs from 24 minutes at Boston to 36 at O’Hare. O’Hare, Atlanta and Denver sit at the top, for different reasons.

Airport	Landing to gate	Door and getting off	Walk	Whole trip, typical
O’Hare	18	10.5	7	36
Atlanta	8	10.5	17	35
Denver	8	10.5	16	35
Boston	7	10.5	6	24
Midway	4	10.5	7	22

O’Hare’s time goes on the ground. Atlanta’s and Denver’s go on the walk and the train. Each stage above is that stage’s own median, so the columns don’t add up to the last one.

For a passenger whose row is fixed, the taxi is the largest source of uncertainty at all 12 airports. It accounts for 53% of the spread in trip time at LAX and 91% at O’Hare. So once a plane is at the gate, the remaining time is much easier to call. If you’re picking someone up, the moment to trust is “at the gate”, and “landed” comes with a wide range.

The gate can matter as much as the airport

Atlanta’s airport-wide figure hides two very different trips. From Concourse F the typical trip is 25 minutes. From Concourse E it’s 40, because the train is the only way to baggage claim.

LAX has the same split: 25 minutes from Terminal 1 and 42 from the Midfield Satellite. Seattle runs 27 from the B gates and 35 from the S gates. JFK runs 26 from Terminal 1 and 33 from Terminal 4.

Row matters less than either. At O’Hare, moving from row 25 to row 15 takes about 3 minutes off.

A checked bag costs less where the walk is long

With a checked bag, this model adds about 8 minutes at O’Hare and about 9 at DFW, but only about 3 at Atlanta and Denver. On a long walk the bag has time to reach the belt before you do. The bag times are judgement, because nobody in the US publishes a measurement, so treat this as a shape and not a figure.

How this was made

For each airport I ran 40,000 simulated trips. Each trip adds up these stages, and the last one applies only with a checked bag.

Stage	How it’s set	Basis
Landing to gate	Drawn from that airport’s real year of flights	Measured
Gate to door open	0 to 2 minutes, most often 1	Judgement
Getting off, row 25	The deplaning model, 737-800, about 82% full	Modelled
Gate to curb	A random gate, then a walk from its mapped route	Modelled
Your bag reaching the belt	A typical figure by airport type	Judgement

The walks come from mapped walkways, in bands that run from a typical walker to a slow one, with trains counted where they’re required. The results above round to whole minutes, because anything finer would claim more than this can show.

What this can’t tell you

Nobody has timed these trips. Nothing here has been compared with a real walk from gate to curb. Published airport walking times run about 1.4 times the typical walker in this model, and the slow end of each band was raised to allow for that.

Each gate counts the same, because public data doesn’t say which gates are busiest. An airport’s figure will shift if your airline uses its far gates. A known gate beats the airport figure every time, so use the airport pages where you can.

It stops at the curb. Rideshare lots, shuttles and rental cars add their own minutes, and at some airports the rideshare pickup is a long way from the curb.

It covers domestic arrivals. A flight from abroad adds passport control, and the landing-to-gate records used here don’t include those flights.

Some maps are thinner than others. Orlando’s walks and the Las Vegas D gates are closer to straight-line distances than routed walks.

It doesn’t cover wheelchair assistance, buses to remote stands or a stop on the way out.

Priority Pass opens 30 of the 148 lounges at the 12 busiest US airports

Sat, 07 Nov 2026 00:00:00 GMT

Priority Pass opens 30 of the 148 lounges at the 12 busiest US airports. At Denver and Newark it opens none.

Lounge finders list what exists. What you need to know is which of those doors opens for the card in your wallet, in your terminal, at your hour.

Two airports, no door

Denver has 8 lounges and Newark 6. Priority Pass opens none of them at either airport.

At the airports where it does open a door, the door is often in one place. Atlanta has 13 lounges and Priority Pass opens 2, both in the International Terminal. O’Hare has 14, and it opens 2, both in Terminal 5. LAX has 17, and it opens 1, with an extra fee at the door.

Orlando is the best share, at 3 of 5. JFK has the most, at 8 of 19, all in Terminals 1 and 4. A lounge in another terminal can mean a second trip through security. At Boston, Terminal A isn’t connected to the others past the checkpoint.

Hours narrow it further

Some of the 30 come with a clock. At O’Hare, the Air France lounge in Terminal 5 doesn’t admit Priority Pass from noon to 5 PM. At Boston the Air France lounge admits it only from 8:00 to 13:30.

The card decides more than the membership

Amex Platinum reaches 60 of the 148, because it stacks several networks: its own lounges, Priority Pass, and Sky Clubs when you fly Delta. Chase Sapphire Reserve reaches 35 and Capital One Venture X reaches 34.

That count flatters every card with an airline condition. Part of the Platinum’s 60 is closed unless you’re on a Delta flight, so check the condition against the flight you’ve booked, and check it on the day.

How this was made

I read 161 airports’ lounges from airport, airline, card and lounge-network pages in November 2026, and restated each fact in plain words. Of those airports, 82 have no lounge at all. The rest hold 422 lounges, and Priority Pass opens 123 of them.

A lounge counts for a membership if any route in its access rules matches. A route that couldn’t be sourced is left out.

The counts here are measured, in the sense that a person read each page and counted. No figure on this page is modelled, and the only judgement is whether a page said what it seemed to say.

What this can’t tell you

Some pages refused to load, and some big airlines’ lounge pages were among them. Of the 148 lounges, 122 are confirmed on a page I could read. LAX is the thinnest, at 7 of 17. A missing route isn’t a “no”, so a card may open more than this says.

United Polaris lounges have no route here, because no readable page states who gets in.

Hours are as the lounge lists them, and they change. Guest rules, visit limits and fees sit outside this count.

Access terms change often, so check the lounge’s own page before you count on one. Airline, card and lounge names belong to their owners.

At Orlando's Terminal C, 6 in 10 arrivals wait over 30 minutes at passport control

Sat, 07 Nov 2026 00:00:00 GMT

At Orlando’s Terminal C, 61.6% of arriving passengers waited more than 30 minutes at passport control over the past year. At Atlanta it was 3.1%.

Those are shares of people, from a year of hourly counts covering 108.7 million arrivals at 46 US airports. Most wait figures average the hours, which treats a quiet hour the same as a packed one. This counts the passengers.

The hall matters more than the airport

JFK has the widest spread at one airport. At Terminal 1, 36.3% waited over 30 minutes and 13.0% waited over an hour. At Terminal 5 those numbers were 10.3% and 1.2%. Terminal 4 sat at 20.2% and Terminal 8 at 26.3%.

Orlando’s gap is wider still. Terminal C ran 61.6% over 30 minutes, Airside 4 ran 14.9%.

LAX shows it too: 20.9% at Tom Bradley, 7.2% at the Satellite 7 hall. Newark’s Terminal B ran 19.6% against 9.0% at Terminal C.

You can’t choose a hall on the day. You can choose the airline that lands there when you book.

Atlanta stands out at the other end. It handled 6.3 million arrivals and only 3.1% waited over 30 minutes.

Late afternoon beats morning, even with twice the people

Across the 13 busiest gateways, the 10 AM hour had the highest share over 30 minutes, at 25.9%. The 6 PM hour had the lowest of any busy hour, at 14.3%, while handling about twice as many passengers.

Single airports show it more sharply. At Boston the share is 32% in the 9 AM hour and 4% at noon. At Newark it’s 7% in the 6 PM hour and 30% at 10 PM, so the pattern isn’t simple and each airport has its own day. The airport pages show it hall by hall.

The season moves it a lot

Across those 13 gateways, December was the worst month at 25.0% and October among the best at 16.4%. Orlando went from 35.1% in February to 55.0% in August. Seattle swung from 37.5% in August to 15.8% in October.

How this was made

Public wait-time reports give each arrival hall’s passengers for each hour, sorted into wait bands: 0 to 15 minutes, 16 to 30, and so on past two hours. I used 205,337 hall-hours from November 1 2025 to November 4 2026, and added up passengers, so a busy hour counts for more than a quiet one.

Every share on this page is measured, as reported. Airport figures for US and visiting passengers come from the same reports as a median of hourly averages. At JFK that’s 11 minutes for US passports and 23 for others. At Orlando it’s 18 and 29. At Atlanta it’s 5 and 8.

What this can’t tell you

The band counts mix US citizens, visitors and Global Entry together, so a hall with more visitors looks slower for that reason alone. Across the 26 gateways with over 500,000 arrivals, the visitor share and the over-30 share correlate at 0.68. Two-thirds of Orlando’s arrivals hold a foreign passport, against 29% at Atlanta. So part of each gap is who lands there, and this page can’t say how much.

It can’t tell you when the clock starts. The reports don’t say where in the hall a wait begins, so the walk from the gate and the wait for your bag are extra.

It only covers airports that report. Flights from airports with US preclearance skip this stage entirely, so they aren’t in these counts.

The past year includes its bad days and its good ones, so this describes the whole year. It isn’t a forecast for your flight. It also isn’t live, and it isn’t a promise.

Hour patterns mix airports. Morning arrivals land at a different set of airports from evening ones, so the 13-gateway line is partly a change in where people land.

Tuesday is the quiet day at airport security

Sat, 07 Nov 2026 00:00:00 GMT

At the 12 busiest US airports, Tuesday has 1 hour a week in the busiest security band. Friday has 54.

That count comes from six months of hourly screening numbers, reduced to a typical week. These are typical bands on a normal day. They aren’t live, and they aren’t a guarantee.

Tuesday is the quiet day

Across those 12 airports the typical week holds 240 airport-hours in the top band, where one airport-hour is one airport for one hour of the week. Friday has 54, Thursday and Sunday 52 each, Monday 51. Saturday drops to 22, Wednesday to 8. Tuesday has 1.

The shape holds further down. Thirty large airports can reach the 20–30 minute band or higher, and they spend 360 airport-hours there on Sunday against 195 on Tuesday. Saturday also comes in at 195 on that count, so Saturday is a quiet day at most large airports and a busier one at the biggest.

If you can choose your day, choose Tuesday. Wednesday is next.

The busy hour depends on the airport

The 6 and 7 AM hours hold 71 of the 240 top-band hours, so the morning advice is right as far as it goes. It doesn’t hold everywhere.

JFK’s busiest hours of the week are 3 to 5 PM, when the afternoon international flights leave. Newark has two peaks, 6 AM and 3 PM. San Francisco’s busiest hour is 10 AM. Denver’s are 8 and 9 AM, later than most. Las Vegas has its busiest hour at 11 AM, and it has a whole day of its own: 8 of its 17 top-band hours fall on Sunday.

Most hours are quiet. Across all 426 airports, 73.2% of the hours in a typical week sit in the “under 10 min” band.

An hour either way often moves the band

At the 12 airports there are 908 busy hours a week, meaning 20 minutes or more in a typical week. In 47% of them, the hour before or the hour after sits in a lower band. In 63%, the same hour on another day of the week sits lower.

So a flight time you can move by an hour is worth moving, and a day you can move is worth more.

How this was made

The raw input is public counts of people screened, by airport, checkpoint and hour. I used the 26 weeks from May 3 to October 31 2026 and left out the 3 weeks holding a major holiday, which leaves 23.

For each airport, weekday and hour, I took the median across those weeks. Then I divided it by that airport’s own busy hour, its 95th percentile. That gives a share of “as busy as this airport gets”. Under 35% is “under 10 min”, under 65% is “10–20 min”, under 90% is “20–30 min”, and anything above is “30+ min”.

Figure	Basis
People screened per hour	Measured
How busy an hour is for that airport	Modelled
Minutes attached to each band	Judgement
The size limits below	Judgement

What this can’t tell you

It can’t tell you how long the line is. The counts measure people who got through, and throughput is capped by open lanes, so a line that doubles can look the same as one that holds steady.

It can’t compare small airports with big ones. Only airports whose busy hour tops 5,000 people can show “30+ min”, and exactly 12 do. Airports above 2,500 stop at “20–30 min”. Everything else stops at “10–20 min”. That rule is a judgement. It means the list of 12 is set by size, so this page can’t claim the other 414 airports never reach 30 minutes.

It can’t see a bad day. Holiday peaks, storms and staffing gaps don’t survive a median.

The hour before the morning peak looks better here than it may feel. Lines often build while lanes are still opening, before the counts rise.

There’s no split for expedited lanes, and no split by checkpoint.

The minutes are coarse on purpose. The best public test I found of this kind of pattern model, published by another company, missed by about 11 minutes on average, and by about 20 at O’Hare. Nothing finer than a band is defensible.

Typical security times are estimated from public passenger-throughput data. Not affiliated with any airport, airline or government agency.

Going south, sit on the right for sunset. Going west, check the date

Sat, 07 Nov 2026 00:00:00 GMT

On a flight heading south, the sunset is on the right every month of the year. Heading north it’s on the left. On an east-west flight the answer changes with the date, and for parts of the year neither side is any good.

These are modelled from sun positions and straight great-circle routes, for seven routes on the 15th of each month.

North and south never change

Minneapolis to Dallas puts the sun off the right window in all 12 months. So does Chicago to Miami. Turn around and it’s the left window, all year. The sun sets in the west, you’re flying across that line, and the season only moves it forward or back along the side of the plane.

The old folk rule works here.

West flips with the seasons

New York to Los Angeles has the sunset off the right side from March to September. In June it sits about 49 degrees to the right of the nose, which is a fine view from a right window. In December and January it moves to the left, but only 13 to 19 degrees off the nose.

Around October and February it sits nearly dead ahead, 5 or 6 degrees off the nose. You won’t see much from either side, because the view is through the cockpit windows.

Two worked examples, both out of JFK:

Leave at 8:00 PM on June 15, and the sun sets off the right side about 2 hours 15 minutes in, 46 degrees off the nose.
Leave at 4:00 PM on December 15, and it sets off the left side about 2 hours 45 minutes in, 13 degrees off the nose.

East puts the sun behind you

Los Angeles to New York has the sunset behind the plane all year. In June it’s 45 degrees behind the left wing. In December it’s about 16 degrees off the tail, on the right. Book the side the chart names if you want the colour in the sky, but you won’t watch the sun go down.

If the sunset is the point of the trip, fly west.

West also makes it last

Flying west, you stay with the sun. On New York to Los Angeles, the longest stretch with the sun low but still up runs 160 minutes in March and 222 in December, for the best-timed departure. Going the other way it’s 31 and 39 minutes. On the ground in Chicago it’s 37 and 45.

Minneapolis to Dallas sits between them, at 66 and 86 minutes.

How this was made

For each route and month, I stepped through departure times every 15 minutes, flew each one along a great circle at 470 knots, and found the minute the sun dropped below the horizon as seen from 35,000 feet. That horizon sits about 3 degrees below level, so sunset comes later than on the ground.

A month’s result is the median across every departure that sees sunset in cruise. “Near the nose” means within 15 degrees, and “near the tail” means within 30.

Every figure on this page is modelled. Sun positions are well known, and the assumptions are the route, the speed and the altitude.

What this can’t tell you

Real flights don’t follow great circles. Winds, air traffic and airspace can bend a route by enough to change the answer when the sun sits near the nose or tail. Treat those months as a coin toss.

Clouds aren’t included, and they decide more sunsets than geometry does.

A wing or an engine can block the view. This doesn’t know your seat, your aircraft or which way the plane turns after takeoff.

It covers seven routes. Other routes run at other angles, and a route between northeast and southwest behaves partly like each case above.

One flight in six at O'Hare takes 30 minutes to reach the gate

Sat, 07 Nov 2026 00:00:00 GMT

At O’Hare, 16.2% of flights take 30 minutes or more to get from the runway to the gate. At Midway, 20 miles away, it’s 0.2%.

The median US flight takes 7 minutes from landing to gate, and 2.5% of flights take 30 or more. Trackers say “landed” at touchdown, so that gap is time the person waiting can’t see.

O’Hare is in a class of its own

Among the 40 busiest airports, O’Hare’s median is 18 minutes and its slowest 1 in 10 takes at least 36. The next airport on the 30-minute measure is DFW at 4.0%, then Seattle at 3.7% and LAX at 3.5%.

At O’Hare, 41% of flights take 20 minutes or more. No other big airport passes 13%.

If you’re meeting someone at O’Hare, “landed” means about 18 more minutes before the door even opens. At Midway it means about 4.

Evening is the slow part of O’Hare’s day

Flights landing in the 7 PM hour at O’Hare take a median 23 minutes, and the slowest 1 in 10 take at least 52. In the 11 AM hour those numbers are 15 and 24.

Other airports have a morning problem instead. LAX runs a 27-minute slow tail at 8 AM and 14 in mid-afternoon. San Francisco’s slow tail is 27 at 8 AM and 12 in the evening. Boston’s is 26 at 5 AM and 11 at noon. Overnight arrivals reaching gates still held by parked aircraft is a likely cause, though these records can’t show it.

Winter isn’t O’Hare’s slow season

O’Hare’s slowest month in this year of data was October, with a median of 20 and a slow tail of 40. January and February were the quickest, with a median of 15 and a slow tail of 32 and 30. Summer ran close to October. August was 19 and 39.

Seattle follows the same seasonal shape more plainly: 9 and 17 in January, 12 and 23 in July and August.

That’s one year of data, and each month appears once.

Your airline can matter as much as your airport

At Atlanta, the median Delta flight takes 8 minutes and the slowest 1 in 10 at least 14. Southwest’s are 6 and 10. Frontier’s are 17 and 53, over 20,177 flights. The 43-minute gap between Southwest’s slow tail and Frontier’s is the widest at any of the 30 busiest airports.

Denver shows the same pattern with bigger volumes: Southwest 6 and 9, SkyWest 10 and 23. At DFW, Frontier’s slow tail is 39 against Delta’s 15.

The records don’t name the terminal or the gate, so they can’t say why. Which gates an airline uses, and how far they sit from the runways, is the likely reason.

How this was made

Airlines report landing and gate-arrival times for every flight. I used all 6.9 million domestic flights at 360 airports from September 2025 to August 2026, with cancelled and diverted flights left out.

A median is the middle flight. “Slowest 1 in 10” is the 90th percentile, by nearest rank. Hours are the local hour of landing, and a chart hour needs at least 300 flights in the year.

Every figure on this page is measured, as reported by the airlines. Nothing here is modelled.

What this can’t tell you

It covers US airlines’ domestic flights only. Foreign airlines and flights arriving from abroad aren’t in it, so it says nothing about a long-haul arrival at JFK or LAX.

It names no terminal, runway or gate, so it can’t tell you which part of an airport is slow.

Airlines appear as the airline flying the plane. A flight sold by United and flown by SkyWest counts under SkyWest.

It doesn’t say why a flight was slow. Waiting for a gate, crossing a runway and a long route around the field all look the same.

It can’t forecast one flight. These are distributions over a year, and a storm or a closed runway isn’t visible here.

The newest month runs about seven weeks behind.