FlyerIntel Research
How Far a Late Start Travels
An aircraft whose first flight of the day leaves late is still more likely to leave late on its sixth flight, even against aircraft at the same airport in the same hour.
Published October 12, 2026 · Source period July 2023 – July 2026
Summary
After a first flight that left 45 or more minutes late, an aircraft’s sixth flight of the day left 15 or more minutes late 47% of the time, against 30% after a start on time or early. Measured on 1,169,215 aircraft-days with 6 or more flights, from the tail number on each flight.
Key findings
- 5,312,998 aircraft-days are in the figures; 22.0% had 6 or more flights.
- After a first flight that left 45 or more minutes late, the second flight left 15 or more minutes late 69.3% of the time, against 8.9% after a start on time or early.
- By the sixth flight the gap was still 16.8 points (47.1% against 30.3%).
- Against aircraft of the same airline that began the same day from the same airport in the same hour and left on time or early, the gap at the sixth flight was 14.6 points.
- Holding the airline and the scheduled hour fixed, the flight number of the day made no more than 0.7 points of difference to how often a flight left late.
- Airlines differ: after a bad start the fourth flight was 21.5 points more likely to be late than after a good one at Envoy Air, and 45.4 at Southwest Airlines.
What was measured
The DOT file names the aircraft that flew each flight. Taking one aircraft’s flights on one scheduled departure date, in scheduled order, gives its day. 5,312,998 aircraft-days with 21,365,348 flights are in the figures. 22.0% of the days had 6 or more flights, and those 1,169,215 days are followed from the first flight to the sixth. For each, the report asks how the first flight left, and how often each later flight left 15 or more minutes late.
How the rest of the day went, by how the first flight left
| First flight | Aircraft-days | Second flight late | Third flight late | Fourth flight late | Fifth flight late | Sixth flight late |
|---|---|---|---|---|---|---|
| Early | 919,885 | 7.9% | 13.7% | 20.7% | 26.3% | 29.0% |
| On time | 173,927 | 14.2% | 20.8% | 28.7% | 34.6% | 36.9% |
| 15 to 44 minutes late | 44,070 | 39.8% | 35.0% | 35.8% | 39.0% | 39.6% |
| 45 to 89 minutes late | 17,953 | 71.0% | 59.6% | 49.6% | 47.9% | 44.2% |
| 90 minutes or more late | 13,380 | 67.2% | 68.9% | 59.2% | 57.3% | 51.0% |
Aircraft-days with at least 6 flights. Each cell is the share of that flight that left 15 or more minutes late. A dash is a cell under 120 flights.
Points more likely to leave 15+ minutes late after a start 45 or more minutes late, against a start on time or early
How far a late start reaches
After a first flight that left 45 or more minutes late, the second flight left 15 or more minutes late 69.3% of the time, against 8.9% after a first flight that left on time or early, a gap of 60.4 points. By the sixth flight the gap was 16.8 points (47.1% against 30.3%), and the average departure delay was 23.0 minutes higher. Aircraft whose first flight left 90 minutes or more late averaged 177 minutes late on it and 52 minutes late on the sixth, while aircraft whose first flight left early averaged 19 minutes late on the sixth. The gap shrinks with every flight and is still 16.8 points at the sixth.
Same airport, same hour, same day
A late first flight can be the airport’s bad day and not the aircraft’s. To separate the two, each aircraft whose day started 45 or more minutes late was compared with aircraft of the same airline that began the same day from the same airport in the same hour and left on time or early (8,181 flights at each position of the day). On the sixth flight the late starts left 15 or more minutes late 50.4% of the time and the matched on-time starts 35.8%, a gap of 14.6 points. That is 87% of the 16.8 points found without matching, so most of what a late start carries to the sixth flight goes with the aircraft and the rest is shared by what the airport and the hour had in common. The matched days are the ones when late and on-time starts were both available at one airport in one hour, so they are weighted toward airports with many flights in an hour, and the gap there is not the gap everywhere.
Against aircraft that left the same airport in the same hour on the same day on time or early
| Flight of the day | Flights after a late start | Left late | Left late after the matched on-time starts | Gap, points |
|---|---|---|---|---|
| Second | 8,181 | 76.1% | 12.0% | +64.1 |
| Third | 8,180 | 68.6% | 19.1% | +49.5 |
| Fourth | 8,174 | 58.8% | 27.0% | +31.8 |
| Fifth | 8,170 | 55.3% | 33.4% | +21.9 |
| Sixth | 8,157 | 50.4% | 35.8% | +14.6 |
Each aircraft whose day started 45 or more minutes late is compared with the other aircraft of the same airline that began the same day from the same airport in the same scheduled hour and left on time or early. Where several were available the comparison is their average, flight by flight. Days with no such pair are left out.
The flight number of the day adds little to the clock
Later flights of the day leave late more often. The sixth flight leaves late more often than the first because it leaves later in the day, when the whole system is further behind, and the question is whether the count of flights adds anything beyond that. Holding the airline and the scheduled hour fixed, no flight number from the first to the sixth was more than 0.7 points from what the airline’s flights at that hour show on average. A flight leaves late about as often at a given hour whether the aircraft is flying its second flight or its sixth. A late start is a different matter: it tells you about that aircraft on that day, whatever the hour.
The flight number of the day against the clock
| Flight of the day | Flights | Left late | Expected from the airline and the hour | Gap, points |
|---|---|---|---|---|
| First | 5,299,967 | 11.9% | 11.7% | +0.2 |
| Second | 4,894,611 | 18.6% | 18.7% | -0.1 |
| Third | 4,170,916 | 23.1% | 23.7% | -0.6 |
| Fourth | 3,303,057 | 27.4% | 27.7% | -0.3 |
| Fifth | 1,983,415 | 30.6% | 30.0% | +0.6 |
| Sixth | 1,167,522 | 31.1% | 30.4% | +0.7 |
| Seventh | 322,431 | 34.1% | 31.8% | +2.3 |
| 8 or more | 166,224 | 28.1% | 25.9% | +2.1 |
All aircraft-days, of any length. The expectation is the share of the airline’s flights that left 15 or more minutes late in the same scheduled hour, over every flight number of the day. An airline-and-hour cell under 120 flights is left out before anything is summed.
Airlines: how long a bad start lasts
| Airline | Months reported | Aircraft-days | Flights per aircraft-day | Days with 6+ flights | Fourth flight late after a start 45 or more minutes late | Fourth flight late after a start on time or early | Difference, points |
|---|---|---|---|---|---|---|---|
| Allegiant Air | 37 | 95,718 | 3.9 | 21.4% | 84.8% | 30.2% | +54.7 |
| Southwest Airlines | 37 | 850,975 | 5.1 | 41.5% | 75.0% | 29.6% | +45.4 |
| JetBlue | 37 | 237,895 | 3.1 | 5.4% | 72.7% | 28.5% | +44.2 |
| Frontier Airlines | 37 | 149,361 | 4.1 | 19.2% | 67.1% | 27.2% | +39.9 |
| Alaska Airlines | 37 | 240,796 | 3.3 | 5.7% | 52.0% | 16.1% | +35.9 |
| Delta Air Lines | 37 | 834,812 | 3.7 | 14.4% | 49.9% | 17.1% | +32.8 |
| PSA Airlines | 37 | 129,024 | 5.4 | 57.6% | 52.2% | 19.5% | +32.7 |
| SkyWest Airlines | 37 | 479,128 | 5.0 | 47.5% | 48.9% | 16.3% | +32.6 |
| Republic Airways | 37 | 210,592 | 4.6 | 34.9% | 43.6% | 14.0% | +29.6 |
| United Airlines | 37 | 781,206 | 3.1 | 2.2% | 49.8% | 23.1% | +26.7 |
| American Airlines | 37 | 866,822 | 3.4 | 9.2% | 52.9% | 28.6% | +24.4 |
| Envoy Air | 37 | 169,719 | 5.1 | 51.2% | 37.6% | 16.1% | +21.5 |
| Spirit Airlines | 35 | 156,385 | 4.0 | 14.7% | 61.9% | 22.3% | — |
| Endeavor Air | 18 | 62,010 | 4.8 | 38.8% | 45.0% | 12.1% | — |
| Hawaiian Airlines | 30 | 48,555 | 4.1 | 28.9% | 53.1% | 7.0% | — |
The two late-share columns count aircraft-days with at least 6 flights and are shown with at least 120 flights. Airlines that did not report every month are shown with their months and are not compared.
Airlines differ in how long a bad start lasts
Aircraft-days run from 3.1 flights on average at United Airlines to 5.4 at PSA Airlines, and 57.6% of PSA Airlines’s days had 6 or more flights. After a start 45 or more minutes late, the fourth flight left 15 or more minutes late between 37.6% of the time at Envoy Air and 75.0% at Southwest Airlines. Measured against the same airline’s aircraft that started on time or early, the difference runs from 21.5 points at Envoy Air to 45.4 at Southwest Airlines. Southwest Airlines, Frontier Airlines and Delta Air Lines carried a bad start the furthest. Envoy Air, American Airlines and Republic Airways carried it the least. The difference is taken against each airline’s own on-time starts at the same flight, so it is not the airline’s overall reliability, and the clear cases are those at least 2 standard errors from no difference with 500 or more flights behind them. Spirit Airlines, Endeavor Air and Hawaiian Airlines did not report every month, so they are listed with the months reported and left out of the comparison.
What this does not show
The aircraft-day is the flights one tail number operated on one scheduled date, so a late-night flight that leaves after midnight belongs to the next date and begins another day. A flight an airline cancelled is not in any figure, so a first flight that never left leaves the day to start with the next one, which may itself be late for the same reason. The DOT records the tail number of the aircraft that flew each flight, so a substitute aircraft starts its own day. 2,467 aircraft-days with two flights in the same scheduled minute, where the tail number is wrong on one of them, are left out. A late start is a symptom as well as a cause: a thunderstorm that delays the first flight is often still there at the fourth, which is why the comparison with aircraft at the same airport in the same hour is shown. Crew, gate and maintenance schedules are not in the data and are a likely part of what the figures carry.
Methodology
Calculated by FlyerIntel from 21,708,321 individual flight records in the US Department of Transportation Bureau of Transportation Statistics Reporting Carrier On-Time Performance dataset, covering July 2023 through July 2026. Following BTS methodology, a flight counts as delayed when it arrives 15 or more minutes after its scheduled arrival time. Cancelled and diverted flights have no arrival time and are excluded from on-time rates rather than counted as late; cancellation rates are reported separately. Full definitions are published at /methodology/. This report uses the tail number the DOT records for each flight. An aircraft-day is the operated flights of one tail number on one scheduled departure date, ordered by scheduled departure, then scheduled arrival, flight number and origin. A day with two flights in the same scheduled minute is dropped. A flight counts as late when it left 15 or more minutes late; diverted flights keep their place in the order and are left out of every rate, and so are flights with no departure delay. The first flight of the day is banded by its departure delay (early; on time; 15 to 44 minutes late; 45 to 89 minutes late; 90 minutes or more late). A start 45 or more minutes late is compared with a start on time or early. The matched comparison groups the days by airline, date, first-flight airport and first-flight scheduled hour, and for each late start takes the average of the on-time starts in its group. The clock comparison gives every flight the late share of its airline’s flights in the same scheduled hour, over all flight numbers. Every cell is printed only when it holds at least 120 flights, and thin airline-and-hour cells are dropped before anything is summed. An airline is called clearly different only above 500 flights and 2 standard errors. Airlines that did not report every month are shown and never compared. Each headline figure was re-derived from the raw flights table by separately written SQL before this report was written.
Limitations
The source dataset covers scheduled domestic flights operated by carriers above the DOT reporting threshold; smaller carriers and international flights are not included. Flights sold under a mainline brand but flown by a regional carrier are attributed to the operating carrier, which is how the DOT records them. Past performance describes what has happened and is not a prediction about any individual flight. The DOT records no crew, gate or maintenance schedule, so the report says how far a late start reaches and cannot say what carries it. Cancelled flights are not counted, and a substitute aircraft starts its own day. The matched comparison covers the days when late and on-time starts were both available at one airport in one hour. The differences between airlines describe the flights each one operated and not how well it is run.
Sources
- FlyerIntel analysis — FlyerIntel · source
- Reporting Carrier On-Time Performance — US Department of Transportation, Bureau of Transportation Statistics · source
Full FlyerIntel methodology · Report a problem with this analysis