The 15-minute rule: what "on time" officially means

A flight that pulls into the gate 14 minutes behind schedule is recorded as on time. The same flight, four minutes slower, is recorded as delayed. Nothing about the passenger experience changes at that boundary, but every published statistic does.

That threshold is the US Department of Transportation's, it dates to the 1980s, and it underpins the 75.9% national figure quoted on this site and the numbers in every airline's own reporting.

The precise definition

A flight is on time when it arrives at the gate fewer than 15 minutes after its scheduled gate arrival time. Three details in that sentence do real work.

Arrival, not departure. A flight can push back 40 minutes late, make the time up in the air, and count as on time. This site reports arrival performance because that is what a passenger is actually buying.

Gate arrival, not touchdown. The clock stops when the parking brake is set at the gate, not when the wheels hit. A flight that lands on schedule and then waits 25 minutes for an occupied gate is a delayed flight.

Scheduled, not scheduled when you booked. The comparison is against the schedule filed for that day. This matters more than it sounds.

The schedule padding problem

Airlines set their own scheduled block times, and they are measured against them. Add fifteen minutes to a route's published flight time and, all else equal, the on-time rate improves without a single operational change.

This has happened, systematically, over decades. A transcontinental flight scheduled at five hours in 1995 is often scheduled at five hours and forty minutes now, on faster aircraft. Some of that is genuine: taxi times at congested airports really have grown. Some of it is a statistic being managed.

The practical consequence for a traveller is that on-time rates are best used to compare, not to predict. Two carriers on the same route are padding against the same taxi times and the same airspace, so the comparison is fair. What the number cannot tell you is whether the flight will be quick, only whether it will be as slow as promised.

Average delay length is the useful companion figure, and this site reports it on every page. Nationally, delayed flights are late by about 53 minutes. A route where flights are rarely late but hugely late when they are is a different proposition from one where they are often late by twenty minutes.

What the threshold hides at both ends

The 15-minute cut throws away information in both directions.

At the good end, it treats a flight arriving 20 minutes early and one arriving 14 minutes late as identical. Nationally the median arrival is around 0 minutes against schedule, which is to say that a clear majority of US flights land early. The on-time rate is decided almost entirely by a tail of badly late flights, not by the typical flight.

At the bad end, it treats a 16-minute delay and a four-hour delay as the same event. This is the more serious loss. Nationally, 8.6% of arrivals land an hour or more behind schedule, and those are the ones that break connections, cost hotel nights, and end trips. This site reports that figure separately for exactly that reason.

Cancelled flights are not late flights

A cancelled flight has no arrival time, so it appears in no on-time calculation anywhere. It is simply absent.

The consequence is uncomfortable: an airline can improve its published on-time rate by cancelling the flights most likely to arrive late. This is not a hypothetical incentive. It is why any honest reading of a punctuality figure has to sit next to a cancellation figure, and why the grade on this site carries a modifier driven by cancellation rate rather than a letter driven by punctuality alone.

Diversions are treated the same way. A flight that ends up at a different airport did not arrive, on time or otherwise.

What the rule does well

For all of that, the 15-minute standard has one large virtue: it has barely changed in forty years, and every US carrier is measured against it identically. That makes the numbers comparable across airlines, across airports, and across decades in a way that a more sophisticated metric redesigned every few years would not be.

It is a blunt instrument, consistently applied. Read alongside cancellation rate, average delay length, and the share of arrivals over an hour late, it supports real conclusions. Read alone, it flatters operations that cancel and pad.

Every page on this site prints all four figures together, and the methodology page states the thresholds verbatim.

Departure delay and arrival delay are different numbers

This site reports arrival performance, and the choice is worth explaining because departure figures are the ones airlines quote more often.

A flight that pushes back twenty minutes late and lands on schedule is, to a passenger, an on-time flight. Nothing was lost. Measured on departure it is a failure; measured on arrival it is a success, and arrival is the honest measure of whether the airline delivered what it sold.

The gap between the two figures is itself informative. Where arrival performance is much better than departure performance, the schedule contains enough padding to absorb ordinary problems. Where the two are close, there is no slack, and every minute lost at the gate is a minute lost at the destination.

The practical reading: do not be alarmed by a late pushback on a long flight, and do be alarmed by one on a short hop, where there is no cruise time in which to recover it.

---

See the definition applied: American Airlines on-time record · Atlanta (ATL) delay statistics · How to read the grades