Taxi Time: The Ground Tax on Every Flight

Summary

Across 14,112,807 flights between the 64 airports FlyerIntel covers, the average scheduled trip spends 27.0 minutes taxiing: 18.3 minutes from the gate to the runway at the origin and 8.7 minutes from the runway to the gate at the destination. That is 16.6% of scheduled gate-to-gate time spent on the ground rather than in the air. On flights under 500 miles it is 30.1%: 26.2 minutes of taxi inside a scheduled trip of 87.2 minutes. And it is not evenly charged: a departure from JFK taxis 27.3 minutes on average against 12.5 at STL, a gap of 14.8 minutes that appears on no ticket.

Two numbers on every US flight record describe time spent on the ground with the aircraft under its own power. Taxi-out is the interval between the aircraft leaving the gate and the wheels leaving the runway; taxi-in is between the wheels touching down and the aircraft reaching its arrival gate. Neither is a delay. Both sit inside the scheduled block time an airline publishes, so a flight can taxi for half an hour and still arrive early. That is what makes taxi invisible: it never appears as a problem, it is never shown when you book, and it is subtracted from your day regardless.

An airport charges taxi twice: once to everything that leaves it, and once to everything that lands. Add the two together and you have the exact amount of taxiing a connection through that airport adds to a trip, on top of the nonstop. ORD (Chicago) is the most expensive of the ten busiest airports on that measure at 43.3 minutes; PHX (Phoenix) is the cheapest at 23.4. The 19.9-minute difference between them is time you spend belted into a seat that is not going anywhere, and it is charged before the connection itself has started. The two halves behave differently, too: taxi-out runs from 12.5 minutes at STL to 27.3 at JFK, while taxi-in runs from 4.1 at SJC to 17.8 at ORD. Leaving costs more than arriving at all 64 airports measured; the usual explanation is that a departure has to wait its turn in a queue for the runway and an arrival does not.

Taxi-out is averaged over every departure from the airport, taxi-in over every arrival. The combined column is what a connection through the airport adds to an itinerary.

Weighted by where flights actually are, taxi-out is lowest at 05:00 at 14.8 minutes and highest at 08:00 at 19.9, a difference of 5.1 minutes across the same set of airports. The national curve is gentle because it averages airports with very different days together. At the airport level the swing is much larger: JFK ranges from 18.6 minutes at 05:00 to 34.3 at 19:00, while MYR varies by only 2.0 minutes across its whole day.

Scheduled departure hour in local time. An hour counts only where the airport recorded at least 2,000 departures in it, and an airport appears only where at least twelve hours clear that bar.

Across the 64 covered airports the correlation between an airport's average taxi-out and its on-time arrival rate is -0.15, which is weak. Airports with long taxi times are not markedly less punctual than airports with short ones, which is what you would expect if the taxi is already inside the published schedule. It takes your time rather than making you late, and that is precisely why nobody notices it. The hourly pattern is not proof that congestion alone drives taxi time. Taxi is partly geometry (how far the gate you leave from sits from the runway in use), and the runway in use changes with the wind, which changes through the day and across the seasons. The mix of airlines, gates, aircraft types and destinations also changes hour by hour, and this data cannot separate any of that from queueing. ATL has more scheduled departures than any other airport here and its taxi-out moves only 3.7 minutes between its best and worst hour, while JFK moves 15.7. Raw size is not the explanation. What the figures do establish is narrower and still useful: at these airports, in these hours, aircraft spent these many minutes on the ground.

Routes between covered airports with at least 750 flights, ranked by the share of scheduled gate-to-gate time spent on the ground.

Three things follow. If a short trip has a driving alternative, compare it against the scheduled gate-to-gate time rather than against the flying time, because a large part of the schedule is already a slow drive across an airfield. If you are choosing between two connecting itineraries, the hub with the shorter combined ground time hands you back the difference for free, and at a slow-taxi hub the connection you were sold is shorter than it looks on paper, because taxi-in happens before you can start walking. And if you have a choice of departure hour at one of the airports in the table above, the early bank is usually the one that reaches the runway quickest.