Washington DC has three airports. The data says they are not equivalent.

Three airports serve Washington, and the choice between them is usually made on ground journey alone. The operational records differ enough to be worth a second thought. Over June 2025 to May 2026: Ronald Reagan Washington National 72.8% on time, Washington Dulles International 80.3%, Baltimore/Washington International Thurgood Marshall 77.3%.

Three different airports in every sense

National sits on the Potomac minutes from the centre, and is constrained by design. Slot restrictions cap how many flights operate. A perimeter rule limits how far most of them can go. Short runways and a river approach leave little margin. The result is a schedule of short domestic hops with tight turns, and short flights have the least cruise time in which to absorb a late departure.

Dulles is the long-haul airport, with the most international traffic and the longest average stage length. Longer scheduled block times carry more padding in absolute minutes, which flatters an arrival figure relative to a short-haul field with identical operational quality. It is also the furthest out, and the ground journey is what travellers most consistently underestimate.

Baltimore is the low-cost field, dominated by point-to-point flying. It behaves like a secondary airport in the pattern described in the hub versus secondary guide: spare capacity, shorter taxi, thinner recovery options when something breaks.

Cancellations diverge more than delays

3.77% at National, 1.39% at Dulles, 1.72% at Baltimore.

Cancellation rate is the figure worth weighting for this market specifically, because the region's weather arrives as whole bad days rather than as a steady drag. A snow event closes the region; it does not gently degrade it. An annual average blends three hundred ordinary days with a dozen severe ones and describes neither.

Choosing

If a non-stop exists from only one of the three, take it. This overrides everything below.

If you are connecting onward, compare the hour-or-more-late shares rather than the headline rates: 11.1%, 7.5% and 7.4%. That is the number that breaks a layover.

If the ground journey differs by more than about forty minutes, it dominates. A few points of on-time rate does not repay an extra hour each way.

If you are flying in winter with a fixed date, read the monthly strip on each airport page. The regional storm risk lands on all three at once, but the recovery profiles differ with the schedule depth at each field.

What none of this captures

Frequency. An airport where your carrier operates your route six times daily absorbs a cancellation in hours; twice daily can mean the next day. Count the departures on the specific route before deciding on reliability figures alone.

Each airport's page here carries its own hour-block curve, twelve-month strip, delay-cause split and full carrier table.

Stage length quietly flatters one of these airports

An arrival figure is measured against the airline's own filed schedule, and long flights carry more scheduled block time in absolute minutes. A transcontinental departure that pushes back twenty minutes late has five hours in which to make some of it up. A 45-minute hop to New York has almost none.

Dulles carries the longest average stage length of the three, National by far the shortest. That difference alone moves an on-time rate by a couple of points before anyone's operational quality enters the picture.

The honest way to compare across the three is therefore not the headline number but the specific route: the same city pair, from each field, on each field's own page. Where two of the three serve the same destination, that is the comparison worth making.

The hour, in a slot-controlled market

Nationally the 5am–8am window runs 89.3% on time against 67.7% for 5pm–8pm. Slot-controlled airports concentrate departures into narrow bands by design, which sharpens that curve: when every carrier wants the same slot, the queue is the delay.

Read the hour-block chart on each airport page rather than assuming the national shape. In this market the useful window is sometimes narrower than "morning".

Frequency is the missing column

A cancellation on a route flown eight times daily costs hours. On one flown twice daily it can cost the night. Both routes might sit at an identical cancellation rate.

Count the daily departures your carrier operates on your specific route before letting a two-point difference in on-time rate decide the airport. That number is your recovery capacity and it appears in none of these figures.

Getting there is part of the comparison

National is the only one of the three on the rail network at the airport itself, which removes a variable the other two leave in place. Dulles and Baltimore both add a transfer or a drive, and both are exposed to corridor traffic at exactly the hours most people travel.

That is not a small factor set against a few points of on-time rate. A flight that arrives fifteen minutes late from a field forty minutes closer to your door has still got you home first. Measure the whole journey, on a weekday, at the hour you will actually make it, and let the airport figures break the tie rather than decide it.

Finally, treat any single month's figure here with caution. A regional snow event or a summer convective run can move a month by several points on its own, which is why the twelve-month strip on each page is the honest view and the annual average is the summary.

---

Compare the three: National (DCA) · Dulles (IAD) · Baltimore (BWI)