Same route, same December weather: Delta cancelled 12.9% of its LaGuardia–Chicago flights, American 1.1%

The U.S. Department of Transportation's on-time database for December 2025 — the most recent month posted (file dated 2026-03-20) — records 582,304 scheduled domestic flights by the reporting carriers, of which 9,001 were cancelled, a cancellation rate of 1.55%. Nationally, of those cancelled flights, 55.8% were coded to weather, 34.7% to "Air Carrier," and 9.5% to the National Aviation System. On the LaGuardia (LGA)–Chicago O'Hare (ORD) pair, the carriers flying it diverged sharply. Outbound LGA→ORD: American Airlines operated 176 flights and cancelled 2 (1.1%); Delta operated 155 and cancelled 20 (12.9%); United operated 389 and cancelled 9 (2.3%). Inbound ORD→LGA was nearly identical: American 3 of 176 (1.7%), Delta 20 of 156 (12.8%), United 7 of 387 (1.8%). The carriers' own DOT cause codes differ as sharply as the totals. All five of American's LGA–ORD cancellations were coded "Weather." Of Delta's 40 cancellations across both directions, 30 were coded "Air Carrier" — the category DOT defines as within the airline's control (crew, maintenance, aircraft turns) — 8 were coded weather, and 2 were coded to the aviation system.

The following is analysis, not fact. Weather cancels flights; it does not usually cancel one airline's flights on a runway while sparing another's. American and Delta flew LGA–ORD through the same December weather; the divergence lives in the cause codes, not the sky. The pattern holds across the whole airport, not just one route. Across all 2,583 Delta departures from LaGuardia in December, Delta cancelled 7.43%, and 58.3% of those cancellations were coded to its own operations. Every other carrier with 500 or more LGA departures coded its cancellations overwhelmingly to weather: American cancelled 2.20% with zero carrier-coded cancellations (100% weather); United 2.48% (89.5% weather); Southwest 2.73% (89.3% weather); Spirit 6.32% (91.4% weather). The weather was shared. The carrier-controlled cancellations were, almost entirely, one carrier's. LaGuardia was the concentrated version of a wider December. Among the four network carriers, Delta's system-wide mainline cancellation rate was 2.75% — the highest of the group, against American at 0.49%, United at 0.39%, and Southwest at 0.55%. For a traveler the operational takeaway is narrow but real: on a route flown by more than one airline, the choice of carrier can move cancellation odds by an order of magnitude in a given month, and the government's own coding attributes much of that spread to operations rather than weather. Of the 624 city-pair-directions in December served by two or more carriers (each with 60+ flights), the widest same-route cancellation gap was 11.8 percentage points — and it was this one, LGA–ORD.

The strongest counter is that the cause codes are self-classified. DOT does not adjudicate each entry; "Air Carrier" versus "Weather" reflects the carrier's own judgment at the time. A skeptic can argue Delta codes marginal-weather cancellations as operational where a competitor codes the same conditions as weather — in which case part of the 30-vs-0 split is a coding-culture artifact, not an operational one SPECULATIVE. The data cannot rule this out. Three more cautions. One month is not a trend: December 2025 may capture a specific Delta operational episode at LaGuardia rather than a durable reliability gap. Hub structure matters: LaGuardia is a dense Delta operation, and tightly banked connecting schedules cascade cancellations in ways a carrier flying the route thinner does not face. And the per-route samples are ~150 flights; a dozen cancellations swing the percentage. The honest claim is a large, cause-coded gap in one month on one hub — not a verdict on either airline's year.

This comparison exists only because U.S. law compels it. DOT requires carriers to file a cause for every delayed or cancelled flight, which is what makes an airline-vs-airline, same-route reckoning possible at all; most aviation authorities publish nothing comparable. In the EU, the same weather-versus-airline-control distinction exists — EU261 compensation turns on whether a cancellation was an "extraordinary circumstance" — but it is litigated after the fact, passenger by passenger, rather than published as data anyone can re-pull.

A traveler sees one cancelled flight and a gate agent saying "weather." Reading all 582,304 December flights at once shows what no gate can: on LaGuardia–O'Hare the weather that grounded planes was the same for every airline, yet one carrier accounted for nearly all the cancellations its competitors' passengers never experienced — and its own DOT filings, not ours, code most of them as operational rather than weather. The most useful number in air travel is not a route's cancellation rate; it is the spread between the carriers flying it, and that spread only appears when you read the whole ledger at once.