How Often Flights Leave Early

Summary

Of the 21,380,414 flights that left the gate in the data, 57.9% pushed back before their scheduled time and 8.4% did so by 10 or more minutes. The airline matters a great deal: Allegiant Air left 10 or more minutes early on 25.1% of its departures and Southwest Airlines on 1.7%, and against the other carriers at the same airport in the same hour Southwest Airlines still ran 12.3 points below them. The busiest hubs leave early least, even after allowing for the airlines that fly from them.

For every flight that operated, the Department of Transportation records the minutes between the scheduled departure and the moment the aircraft left the gate. A negative number is an early departure. Five minutes early is routine and costs a passenger nothing, so the headline figure is 10 or more minutes, and the tables give 5 and 15 as well. Most airlines stop boarding some minutes before the scheduled departure, and a flight that pushes back early has finished boarding earlier still. The data does not record boarding or missed passengers, so it cannot say how often anyone was left behind; it says how often the aircraft left the gate ahead of the time on the ticket. Cancelled flights are not counted. There are 21,380,414 departures in all.

Each row is a slice of the departure delay, in whole minutes. The last column is the share that arrived on time among flights that reached their destination, shown when a row holds at least 120 of them.

Among the 12 carriers that reported every month, the share of departures leaving 10 or more minutes early runs from 1.7% at Southwest Airlines to 25.1% at Allegiant Air, a factor of 15. Part of that could be where and when a carrier flies, since flights leave early more often from smaller airports and in quiet hours. To take that out, each carrier’s flights are compared with the other carriers’ flights at the same airport in the same scheduled hour, in cells where both have at least 120 departures. Against that expectation Southwest Airlines’ departures left 10 or more minutes early 12.3 points less often than the other carriers’ flights did, and Frontier Airlines’ 13.4 points more often. The order of the carriers barely changes from the raw figures, so the difference is not explained by where and when a carrier flies. The data shows the difference and not the reason for it.

Ordered by the share leaving 10 or more minutes early. The last column is the carrier’s share leaving 10 or more minutes early minus the share the other carriers achieved in the same airport-and-hour cells, weighted by the carrier’s own departures; it is given only for carriers that reported every month and only when at least 50% of their departures sat in such a cell. Endeavor Air, Spirit Airlines and Hawaiian Airlines reported fewer months than the others, so their figures cover different seasons and are not compared.

Airports differ, and part of the difference is which airlines fly from them. For each covered airport the table gives the share leaving 10 or more minutes early, the share its carrier mix would predict at each carrier’s rate at its other airports, and the difference. The largest airports come out below their expectation: DEN at 4.7% against 8.5% expected, ATL at 3.3% against 9.8% expected, DFW at 3.0% against 8.3% expected, ORD at 3.6% against 9.3% expected and CLT at 2.2% against 9.8% expected. Across the 64 airports the bigger the airport, the further below its expectation it sits (rank correlation -0.83); 31 of 64 are below it. At the other end MYR (Myrtle Beach) left 10 or more minutes early on 23.8% of departures against 10.4% expected. The expectation allows for the airlines and not for anything else about the airport, such as its size, its weather or how its gates are laid out.

Departures of the carriers that reported every month. Expected is the airport’s departures for each carrier multiplied by that carrier’s share leaving 10 or more minutes early at all its other airports. Ordered by the difference.

By scheduled hour, the highest share leaving 10 or more minutes early is 3 AM–4 AM at 30.0% (2,587 departures) and the lowest 4 PM–5 PM at 6.9%. Hours follow the late-departure pattern less closely than weekdays and months do. By weekday it is Tuesday at 10.1% against Sunday at 7.3%, and ranked by the share leaving 15 minutes late the weekdays fall in almost the reverse order (rank correlation -0.96). By calendar month it runs from July at 6.0% to November at 9.9% and the months again fall in nearly the reverse order of late departures (rank correlation -0.88). Across the individual months in the data the range is 4.5% in July 2023 to 11.8% in November 2025. Some of that is arithmetic: when departures run later the whole spread of departure times moves later and fewer flights are left in the early tail. It also fits a system with room to spare letting aircraft go when they are ready, but the data has no measure of room and cannot separate the two.

Weekday of the scheduled departure date. The last column is the DOT’s own line for a late departure.

All years in the data pooled by month of the year. July is in the data four times and the other months three.

The DOT time is when the aircraft left the gate, not when a door closed or a passenger was refused boarding, so the data cannot count anyone who missed a flight that left early. Airlines set their own boarding cutoffs, and none of them is in the data. A flight can leave early because it boarded quickly, because a gate or a slot became free, or because a schedule was padded, and the report cannot tell those apart. The matched comparison removes where and when a carrier flies but not everything about how it flies, such as aircraft size or the length of its routes. The seasonal and weekday patterns mix the weather with traffic and with the size of the schedule.