OnTimeIndex

How to Tell If a Flight Route Is Reliable Before You Book

On-time rate, median vs. 90th-percentile delay, and cancellation rate — what each number on a route actually tells you, and why one good headline figure can hide a bad flight.

A booking site shows you a price, a departure time, and maybe a green “on time” badge lifted from the last few days. None of that tells you whether the route you’re about to book runs late as a rule. That question has a public answer: every U.S. carrier files its actual departure and arrival times with the Department of Transportation, month after month, and the reduced record shows how a route has behaved over years, not days.

Reading it well means understanding four numbers and, more importantly, how they interact. A route can look excellent on one and be a poor bet anyway.

On-time rate is the headline, not the whole story

The on-time arrival rate is the share of flights that reached the gate less than 15 minutes after schedule — the federal 15-minute rule (14 CFR § 234.2), used unchanged. A route that arrives on time 82% of the time is genuinely more reliable than one at 68%. But the rate says nothing about how bad the late flights were, and it can be propped up by cancelling the flights that would otherwise have counted as late. So it’s the number you start with, never the one you stop at.

Median delay tells you the ordinary day; the 90th percentile tells you the bad one

Two routes can share an on-time rate and feel completely different to fly. The median delay is the typical experience — half of flights do better, half worse. The 90th-percentile delay is the tail: one flight in ten is at least this late. A route with a median near zero but a 90th percentile of two hours is one where most days are fine and a real minority are miserable. If your trip has a hard downstream commitment — a connection, a cruise, a wedding — the tail is the number that should drive the decision, not the average.

Why the average alone misleads: flight delays are heavily right-skewed. A handful of three-hour delays drag the mean far above what a typical passenger experiences, which is exactly why the federal record and this site report percentiles instead of a single average.

Cancellations are the number the on-time rate hides

A cancelled flight isn’t a late flight — it drops out of the on-time calculation entirely. That means a carrier facing a bad operating day can improve its published on-time rate by cancelling the flights it would have run hours late. So a route’s cancellation rate belongs right next to its on-time rate: 85% on time sounds strong until you see that 4% of departures never left at all. Diversions — landing somewhere other than the booked destination — are rarer but count the same way against a smooth trip.

Compare carriers on the same route, not routes in the abstract

The most useful reliability question is rarely “is this route good?” but “which airline flies it best?” The same city pair, flown by three carriers, can show a ten-point spread in on-time rate and very different cancellation behaviour, because schedule padding, hub congestion, and aircraft rotation differ by airline. Line the carriers up side by side over the same trailing window and the choice usually makes itself.

The Route Reliability Check does exactly this: enter an origin and destination and it returns every reporting carrier on the route with its on-time rate, median and 90th-percentile delay, and cancellation and diversion rates over the trailing 12 months — each figure reproducible from the public BTS download.

Sources

Informational only — every figure here comes from historical federal filings, not a prediction of any future flight, and nothing here is professional travel advice.