FlyerIntel

How to read an airport’s reliability profile

An airport page on FlyerIntel has more than a dozen sections, and no airport shows all of them. This guide goes through what each section measures and where its figures come from, then sets out how to compare two airports without being misled by size, by the airline that dominates a hub, or by the time of year.

What the numbers describe

Every rate on an airport page describes departures from that airport, judged on when they arrived at their destination. A flight from Atlanta to Chicago that lands late is counted against Atlanta, not Chicago. An airport’s rate therefore reflects the places its flights go as well as the airport itself. The page says this under the reliability heading, and it is the first thing to hold in mind when two airports are compared.

Apart from the live conditions at the top, everything is history over one fixed period. The first section, “Right now”, uses FAA operational events and the current weather observation; when a live source is missing, the page shows the historical record for that hour and labels it as history.

The sections, in page order

Airport reliability opens with the reliability score, from 0 to 100, and its percentile among the covered airports. The score combines six components with fixed weights: on-time arrival rate 35%, cancellation rate 20%, severe delay rate 15%, median arrival delay 10%, average delay when late 10%, worst-case arrival delay 10%. Beside it are the plain figures: the share of departures arriving on time, the cancellation rate, the median arrival delay, the share of arrivals 60 minutes or more late, the average delay when late, and the number of departures behind all of them. The definitions are in the guide to how on-time performance is counted.

What stands out picks out one respect in which the airport differs most from comparable airports, such as how widely its on-time rate swings by time of day. Why flights are cancelled splits cancellations by the four reason codes airlines file, and appears only for an airport with at least 500 coded cancellations. Why flights run late gives the split of delay minutes among the five DOT causes, explained in the guide to why flights are delayed, together with a measure taken from the aircraft itself: of the departures that left 15 minutes or more late, the share that followed a flight the same aircraft had landed 15 minutes or more late.

What weather costs gives the cancellation and on-time rates for each kind of weather at the airport, set against normal conditions, and Which airlines hold up in bad weather does the same for each carrier against the others at that airport in the same months. Turning aircraft around measures the time between one flight and the next flight of the same aircraft, and what a late arrival does to the departure that follows. What sets the airport apart lists the hours, months, airlines and taxi times where it departs most from the other airports.

Best time to fly gives the on-time rate by scheduled local departure hour and window, and the weekly and seasonal patterns give it by weekday and by month. A bucket is shown only with at least 120 flights behind it. What to expect gives departures a day, the busiest hour and typical taxi times, with a chart of scheduled departures by hour. The airlines section lists each carrier with at least 500 departures and its on-time rate, cancellation rate and median delay, the routes section the routes that do best and worst, and Reliability over time charts the monthly on-time rate. Where they are published, parking and international arrivals sections cover costs and customs waits.

A section that is missing is missing because its sample was below the published floor, or because the source did not publish the figure. It does not mean the airport did well. FlyerIntel leaves a heading out rather than print a conclusion drawn from too few flights.

Comparing airports fairly

Check the sample. Over July 2023 – July 2026, covered airports ranged from 11,763 departures at ABE to 1,012,247 at ATL. An on-time rate over 11,763 flights moves further on a few bad weeks than one over 1,012,247. The compare tool marks a gap as clear only when it is larger than chance would produce at about 95% confidence, using a two-proportion test on the flights behind each rate. When the gap is inside that range it still shows which side is higher, but it does not call it.

Look at the airline mix. At a hub, the figures are dominated by the airline that runs it. At 20 of the 64 covered airports, one operating carrier flew more than half of all departures. Two airports can differ because their dominant airlines differ, not because of the runways or the weather. The “Airlines” section of each page shows the rate for each carrier at that airport, and the research on same airport, different airlines shows how far apart carriers at one airport can be.

Compare the same months. Every airport is measured over the same period, but an airport whose busiest months fall in thunderstorm season or in winter will look worse on a full-period figure than one whose year is milder. The monthly and seasonal tables let two airports be compared in the same month. A seasonal figure appears only where at least two years of that month are on file, because one year of one month is weather, not a season. Cancellations are also uneven: a small number of very bad days account for much of an airport’s total, which the report nobody flies on an average day measures.

Choose between airports of one city only on the evidence. Where a city has several airports, the metro area pages put them side by side for a given destination. In most of the cities with more than one covered airport the differences are small, as the report on same city, different airport found.

For scale: over July 2023 – July 2026, the highest on-time rate among the covered airports was HNL at 85.3% and the lowest was FLL at 72.6%. They sit among 64 airports that differ in size, airline and climate, so they describe the range and not a verdict on either one.

Looking up your own case

Start from the airport index or the reliability map, open the airport you will use, and read the sections above in this order: the score and the cancellation rate first, then the delay causes, then the best-time and weekday tables for the day you plan to fly. The rankings show where it sits among the rest. The guide to choosing a flight continues from there.

What this is based on

The airport figures quoted here are for the 64 covered airports, July 2023 – July 2026. The thresholds and weights are read from the same configuration the pages use, and are listed on the methodology page.

An error in a figure or a definition can be reported on the corrections page. How FlyerIntel decides what to publish is set out in the editorial policy.