US domestic flights · Jan 2023 – Aug 2026
Why US flights run late
One in five arrivals lands at least 15 minutes late. Three questions about those delays: what causes them, when they happen, and where.
- Arrivals delayed ≥15 min
- 21.1%
- Arrivals cancelled
- 1.6%
- Arrivals analysed
- 27.8M
Source: Bureau of Transportation Statistics, Airline On-Time Performance (delay cause). One record is one airline at one arrival airport in one month. “Delayed” means arriving 15 or more minutes late; cancelled and diverted flights are counted separately. 2026 covers January to August.
1What is the biggest cause of flight delays?
Late aircraft tops both measures, but it’s a knock-on effect. Trace it back and the airlines’ own operations cause most delays.
Late aircraft accounts for 40% of delay minutes overall, but how much depends on the airline. Airlines that turn planes around quickly let one morning delay cascade through the rest of the day.
2Which time of year is busiest, and does busy mean delayed?
Flight volume barely changes through the year, but delays nearly double from fall to summer. Summer is the worst season, not the holidays.
If busy meant delayed, the busiest months would also be the worst. They aren’t: volume and delay rate are only loosely related, and the gap from the Nov low (16%) to the Jul peak (28%) comes mostly from weather and stretched summer schedules.
3Which airports experience the most delays?
Size doesn’t predict delays. San Francisco, Florida and San Juan run late more often than their airlines do elsewhere.
Counting delayed flights just ranks airports by size: the 30 busiest handle 64% of arrivals and 64% of delays, so a count map is really a population map. Every chart below applies at least one of four fixes:
- 1Rate, not count% of arrivals delayed
- 2Volume floordrop or shrink small, noisy airports
- 3Airline-mix adjustmentcompare to what its airlines do nationally
- 4Cause fingerprintairspace limits vs extreme weather
Filters apply to every chart in this section. The floor removes airports with too few arrivals for a stable rate. Shrinking keeps every airport but blends its rate with the US average (21.1%) in proportion to how little data it has: an airport with 171 arrivals moves halfway, while ATL barely moves. The strength (k ≈ 171 arrivals) is estimated from the data. Shrinking adjusts rates (charts 1–3); the cause rankings (chart 4) are unshrunk. Turn both off to see the noise that both fixes deal with.
Some airports share a high delay rate because they share airlines. The next chart compares each airport with what its own airlines achieve nationally in the same month. What remains is the airport’s own effect.