Same route, opposite direction: why the return leg runs differently

Between Seattle (SEA) and Charlotte (CLT), the two directions of one airport pair are practically two different products. SEA to CLT arrived on time 87.3% of the time from September 2025 to August 2026. CLT to SEA managed 67.5%. That is a 19.8-point gap on the same pair of runways, and anyone who books the round trip is buying both numbers.

It is not a freak. Of the 1,190 airport pairs flown at least 1,000 times in each direction, 92 differ by ten points or more between the outbound and the return. The DOT measures each direction as a separate operation, and this site gives each direction its own route page for that reason.

I kept the two directions as separate records from the first step of the pipeline, and this is the payoff. It is why LAX to SFO and SFO to LAX can sit ten points apart on this site instead of meeting in a middle that neither flight delivers.

The busiest pairs, both ways

Among the busiest pairs, most run close to level. The busiest pair, New York and Chicago, carried 23,260 flights in the year. LaGuardia to O'Hare, the core of that market, ran 66.7% on time against 66.6% for the return. Both ends are congested, both suffer in the evening, and the two directions share the same weakness. Only 2 of these 12 pairs differ by five points or more.

The widest split among them is Los Angeles (LAX) and San Francisco (SFO). SFO to LAX arrived on time 70.9% of the time; LAX to SFO managed 60.1%, a gap of 10.8 points.

Why one direction loses

Three mechanisms account for most of the gaps, and a route can suffer from more than one.

The arrival end is metered. Some airports lose arrival capacity in specific conditions, and the FAA holds flights at their origin until there is room. San Francisco is the textbook case. Its parallel runways sit too close together to use side by side for landings in low cloud, so when the marine layer sits over the bay the arrival rate falls and inbound flights wait on the ground at the other end. Los Angeles to San Francisco takes that directly: 56% of its delay minutes are charged to the National Aviation System, the DOT category for air traffic control and non-extreme weather. Its weakest departure window is 8am–11am, at 49.8%, consistent with morning cloud that is slow to clear, and its worst month was May at 44.1%.

The route's whole record is pulled down by a stretch I did not believe at first. From April to July 2026, flights into San Francisco fell to roughly 35 to 50% on time on several routes, and my first assumption was a bug in the aggregation. Checked against the raw month files, it is in the data. The data does not say why.

San Francisco to Los Angeles carries a different signature. Only 20% of its delay minutes are air traffic; 58% are late-arriving aircraft. Aircraft held up on the way into SFO leave late on their next leg. The southbound direction inherits San Francisco's problem second-hand, and diluted.

{
 "flights": 11437,
 "completed": 11339,
 "onTimePct": 70.9,
 "veryLatePct": 9,
 "avgDelayMin": 58.4,
 "medianArrDelay": -2,
 "cancelPct": 0.82,
 "divertPct": 0.03,
 "causes": {
  "carrier": 0.2107,
  "weather": 0.0095,
  "nas": 0.2037,
  "security": 0.0002,
  "lateAircraft": 0.576
 },
 "byHourBlock": [92.6, 74.7, 60.4, 63.9, 70.2, 59.1 ],
 "byMonth": [86, 77.4, 85.4, 69.6, 65.3, 60.7, 61.7, 57.7, 78.6, 73.6, 74.2, 71.2 ],
 "byDow": [68.4, 77, 73.9, 65.7, 66.3, 77.7, 68.6 ]
}

This is the file the pipeline writes for the southbound route each month; every SFO to LAX figure in this guide is recomputed from it, and none from the northbound record.

The time of day differs each way. Airlines schedule the two directions of a route for different markets, so they rarely leave at mirror-image hours. Transcontinental flying shows the principle: eastbound schedules include red-eyes, which leave into empty night-time airspace, while many westbound flights leave the East Coast during the evening departure push. On the New York to Los Angeles transcon itself the two schedules roughly cancel out, with JFK to Los Angeles at 75.4% and the return at 77.5%. On a long-haul pair like Seattle (SEA) and Charlotte (CLT), where one direction is built around overnight and early-morning departures and the other leaves a hub in the evening, the effect is likely much larger.

The aircraft comes from somewhere. A flight leaving a hub late in the day uses an aircraft that has already flown several legs through a congested system, and it may wait for connecting passengers. The same aircraft's return from a small city is often its first flight of the next morning, after a night parked at the outstation. Nationally 40% of delay minutes are late-arriving aircraft, and on a hub-and-spoke route that chain likely runs mostly one way.

The biggest gaps

The gap ranking is dominated in this period by two kinds of route. The first is short hops into San Francisco (SFO) from elsewhere in California, where the flight toward SFO runs far worse than the flight away from it. The second is flights leaving American's hubs at Dallas-Fort Worth (DFW) and Charlotte (CLT), where the outbound from the hub runs well behind the flight coming in. Both fit the mechanisms above: a metered arrival airport in the first case, late aircraft and evening banks in the second.

Seasonal weather at one end adds a third pattern. Seattle to Anchorage ran 76.5% on time, falling to 54.6% in December, when flights land into Alaskan winter snow and darkness. Anchorage to Seattle ran 86.9% on a schedule that leans on overnight departures into an uncongested Seattle morning.

Short shuttles can split too. Houston Hobby to Dallas Love Field ran 81.6%, the reverse 74.8%. Love Field's departures as a whole run 72.6% on time, falling from 91.5% before 8am to 59.9% between 5pm and 8pm. A short flight is only as good as the airport it leaves.

How I book a round trip now

Look up the return leg on its own page. A route's reputation is usually built from the direction people fly most or complain about most. On this site every route has two pages; open both before booking a round trip.

Put your slack on the weak leg. If the flight home is the weak direction and you have work the next morning, the early departure on that leg is the one worth paying for. At 60.1% on time, I would not put a connection behind the northbound Los Angeles to San Francisco flight. If the outbound is the weak one and something important waits at the destination, go a day early.

For SFO arrivals, look at the other Bay Area fields. Oakland (OAK) and San Jose (SJC) do not share San Francisco's runway geometry. The Bay Area airports guide compares them.

Leaving a hub late in the day, expect the chain. An evening departure from Dallas-Fort Worth or Charlotte to a smaller city carries the whole day's delays with it. The morning flight in the other direction does not. The connection risk tool shows how much buffer a hub needs.

Do not trust a round-trip average. Two directions at 70.9% and 60.1% average to a figure neither flight delivers.

Check both directions: LAX to SFO · SFO to LAX · JFK to LAX · LAX to JFK · Compare any US route