Worst months to fly in the US, ranked by the data

In the current twelve months of DOT records, July was the weakest month for US domestic flying at 71.1% on time. September was the strongest at 83.4%. That 12.3-point spread is larger than the gap between most competing airlines on most routes.

Unlike the airline you pick or the hour you leave, the month is often fixed by the reason for the trip. Where it is not fixed, moving a flexible week is one of the cheapest reliability improvements available.

Two different bad seasons

Summer and winter both produce delays, and they produce them in completely different shapes.

Summer delays are frequent and diffuse. Convective weather in the Southeast, Texas and the Midwest builds through the afternoon almost daily from June through August. A single cell parked over a hub reroutes traffic for two hours, and the effect ripples for the rest of the evening. Add peak leisure demand filling every seat and leaving no rebooking capacity, and July looks like a month of persistent forty-minute delays rather than a month with three disasters in it.

Winter delays are rare and catastrophic. Most days in January are fine. Then a storm system crosses the Northeast, three hub airports stop operating, and thousands of flights cancel over 48 hours. The monthly average conceals both facts: it hides how good the ordinary January day is, and it hides how completely a bad one fails.

The distinction matters for planning. In summer, build buffer into every connection because the risk is spread across every day. In winter, watch the forecast and be ready to move the whole trip, because the risk is concentrated in a handful of identifiable days.

The reliably good stretch

Late September through early November is consistently the strongest window in US flying, and the reasons stack up neatly.

Summer convection has stopped. Winter storms have not started. School holidays have ended, so schedules are lighter and load factors drop, which means an aircraft going out of service does not strand a full cabin with nowhere to rebook. Airlines also trim frequencies in the shoulder season, which reduces congestion at exactly the airports where congestion is the binding constraint.

Late April and May behave similarly for the same reasons in reverse, though spring convection starts earlier in the South.

The holiday exception, and what it actually looks like

Thanksgiving and the last two weeks of December are the two periods travellers fear most, and the data supports the fear with one important qualification.

The problem is not that holiday operations are worse. It is that they have no slack. On an ordinary Tuesday in October, a cancelled flight means a seat on the next one three hours later. On December 23rd, every flight is full, and a cancellation can mean a two-day wait. The delay statistics for the period are often unremarkable; the consequences of landing in the wrong tail of them are severe.

The practical response is not to avoid flying at the holidays. It is to fly on the days everyone else does not, take the first departure of the day so a cancellation still leaves options behind it, and avoid connecting through a hub that is exposed to winter weather when a non-stop or a southern routing exists.

What this means for a specific trip

National monthly averages are a starting point, not an answer. Seasonality is intensely local.

Denver's difficult months are not Miami's. A route into a Florida leisure market peaks in March and collapses under hurricane-season disruption in September, while a Rocky Mountain route has its worst month in January and its best in July. Averaging those together produces a national curve that describes neither.

Every route, airport and airline page on this site carries its own twelve-month strip with the flight count behind each month, which is the version worth reading before booking. The national pattern tells you what to expect if you know nothing else. The route's own strip tells you what to expect on the flight you are actually considering.

The short version

If the dates are flexible, aim for late September through early November. Treat July and August as a period requiring longer connections rather than a period to avoid. Treat late December as a period where the median outcome is fine and the bad outcome is very bad, and book accordingly.

And check the specific route. July being the national low point tells you very little about a route between two cities where July happens to be the quiet season.

Months are the wrong unit for a fixed date

A monthly average is the right tool for choosing between June and October. It is close to useless for deciding whether to fly on the 23rd.

Delay risk within a month is not evenly spread. In summer it clusters in the late afternoon of convective days, which are frequent but individually short. In winter it clusters in whole storm days that shut a region down. In late December it clusters around the four or five heaviest travel days, when the problem is not weather at all but the complete absence of spare seats.

If your dates are fixed, the monthly strip on each page tells you what kind of risk you are exposed to rather than how much. A route with a winter-peaked profile says: watch the forecast, be ready to move the whole trip, and expect most days to be fine. A summer-peaked profile says: the forecast will not help much, so build buffer into every connection instead.

If your dates are flexible, the monthly figure is exactly the right tool, and moving a week is one of the few decisions that reliably buys several points.

---

Check a route's own seasonal pattern: Orlando (MCO) delay statistics · Denver (DEN) delay statistics · Compare any US route