May 2026 · US Department of Transportation filings
Extreme weather: 4.1% of delay minutes.
Carrier delay and late-arriving aircraft: 75.2%.
Every large US airline reports its delays to the federal government each month, including what caused them. Across 442,954 domestic flights in May 2026, 75.2% of reported delay minutes were coded to the carrier's own operation or to an aircraft arriving late from an earlier leg, against 4.1% coded as extreme weather. These are the carriers' own filings, not our estimates, and every figure below links to the source.
75.2%
of delay minutes coded to the carrier or a late inbound aircraft
4.1%
of delay minutes coded to extreme weather
21.9%
of flights arrived over 15 minutes late
Evidence brief · May 2026
Where the filed delay minutes were coded.
Of reported delay minutes, 75.2% were coded as carrier delay or late-arriving aircraft; 4.1% were coded as extreme weather.
32.1%
coded as carrier delay — crew, maintenance, fuelling, baggage
43.1%
coded as late-arriving aircraft, a knock-on from an earlier leg
20.5%
coded to the national airspace system
4.1%
coded as extreme weather
What this does not say
- It does not say 75.2% of delay is the airline's fault. Late-arriving aircraft is a knock-on code: BTS records that the inbound aircraft was late, not what made it late. Some of it began as weather or a traffic-flow restriction earlier in the day, and the filings do not let anyone say how much.
- It does not say weather causes only 4.1% of delay. Weather managed through air-traffic flow control is filed under the national airspace system, so real weather impact is spread across two codes and is higher than the extreme-weather column alone.
- What it does say is that 32.1% of reported delay minutes are coded directly to things inside the carrier's own operation — more than seven times the extreme-weather share — and that is a carrier's own filing, not an outside estimate.
Basis: 442,954 domestic flights and 6,220,554 filed delay minutes across 7 reporting carriers, May 2026. Shares are of reported delay minutes pooled across carriers, not an average of per-carrier percentages.
Ranked by how often they arrive late
Southwest Airlines ran latest at 27.5%; Alaska Airlines was most reliable at 16.7%. Compare late arrivals, cancellations, and the delay-cause categories reported to BTS.
| Airline | Flights | Over 15 min late | Cancelled | Avg delay | Carrier delay + late-arriving aircraft | Weather |
|---|---|---|---|---|---|---|
| Southwest AirlinesWN | 120,646 | 27.5% | 0.4% | 15.6 min | 81.1% | 2.3% |
| Frontier AirlinesF9 | 18,423 | 23.2% | 0.5% | 18.8 min | 75.9% | 2.3% |
| American AirlinesAA | 88,740 | 23% | 1.5% | 20 min | 79% | 7.1% |
| United AirlinesUA | 74,200 | 19.9% | 0.4% | 14.1 min | 62.1% | 3.9% |
| JetBlue AirwaysB6 | 19,806 | 18.5% | 0.2% | 12.7 min | 69.2% | 2.6% |
| Delta Air LinesDL | 91,955 | 17.5% | 1.2% | 12.2 min | 73.7% | 3.2% |
| Alaska AirlinesAS | 29,184 | 16.7% | 0.7% | 9.1 min | 70.5% | 4.2% |
These are monthly aggregate filing codes. They cannot establish what caused your individual flight, whether it was within the carrier's control, or whether compensation is owed.
Where the delay minutes actually go
Each bar is one carrier's delay minutes split by the cause it filed. Red is carrier delay and orange is late-arriving aircraft; blue is the national airspace system and grey is extreme weather.
- Carrier — crew, maintenance, fuelling
- Late inbound aircraft
- Air traffic system
- Weather
Down to the route
A carrier-wide figure hides a lot: the same airline can run 12% late on one city pair and 40% on another. We publish the same breakdown for every busy route in the dataset, each direction measured separately.
Browse route delay recordsMethodology, and what this does not cover
These are monthly aggregate filing codes. They cannot establish what caused your individual flight, whether it was within the carrier's control, or whether compensation is owed.
- Source
- The US DOT Bureau of Transportation Statistics Reporting Carrier On-Time Performance database, month 2026-05, retrieved 2026-07-28. Nothing here is modelled, estimated or crowd-sourced. Airlines file these numbers themselves under federal requirement.
- “Carrier delay + late-arriving aircraft” is our sum, not a BTS field
- BTS reports five causes separately. We add carrier delay to late-arriving-aircraft delay because both reach the flight through the carrier's own rotation rather than through a weather, airspace or security code, but the sum establishes no cause: the late-arriving code records that the inbound aircraft was late and not what delayed it, so an unknown part of it began as weather or traffic flow earlier in the day. Both components are shown separately above so you can reject the sum and use the raw carrier figure instead. It changes the ranking: Delta has the highest raw carrier share but among the lowest knock-on share.
- Minimum sample
- Route and airport figures are withheld below 300 flights in the period. A thin route produces dramatic percentages from a handful of bad days — one route showed 62% late off 106 flights, which is noise, not a finding. Carrier totals need no threshold; the smallest here files over 18,000 flights.
- Domestic US only — no Mexico, no Europe
- 14 CFR part 234 requires reporting from US carriers above 0.5% of domestic scheduled-service passenger revenue, and it covers domestic service. We checked: across 442,954 flights, no Mexican airport appears at all. So this says nothing about any cross-border route, and nothing about Volaris, Viva Aerobus or Aeroméxico. Mexican carrier data has to come from AFAC, which is a separate source we have not published.
- One month, not a trend
- This is a single reporting month and carries that month's seasonality. It is a snapshot, and we would rather label it as one than present a single month as a pattern.
- Reproduce it
- The figures come from the bulk file BTS publishes at a stable URL, aggregated by a script in this repository. Every number on this page is also published as a file with the raw counts it was derived from, so you can check the arithmetic without re-downloading the federal source, and check it against that source if you want to. Our full methodology sets out the sourcing and publication rules behind this and every other figure on the site.
Figures cover May 2026, a single reporting month, from the US DOT Bureau of Transportation Statistics On-Time Performance database. Retrieved 2026-07-28. Reporting under 14 CFR part 234 covers domestic US scheduled service only, so no international route appears — including anything to or from Mexico.
Download the data
The table above as a file, with the raw counts every percentage was derived from, so the arithmetic can be redone rather than taken on trust. Released under CC0 1.0 — no permission needed, no attribution required, no sign-up.
Provenance and reproducibility
- Reporting period:
- May 2026 (2026-05)
- Rows:
- 7
- Methodology version:
- v1.0.0
- Generated:
- 2026-07-28
- Retrieved:
- 2026-07-28
- Source:
- US DOT Bureau of Transportation Statistics — Reporting Carrier On-Time Performance (1987-present)
Reproduce it from the federal source
- Download the Reporting Carrier On-Time Performance file for May 2026 from https://www.transtats.bts.gov/ontime/.
- Keep rows for the reporting carriers listed in this file; no other filtering is applied.
- Group by reporting carrier and sum: flights, arrivals more than 15 minutes late, cancellations, diversions, arrival delay minutes, and the five filed delay-cause minute columns.
- Rates divide by flights. Cause shares divide by the sum of the five cause columns, not by total arrival delay minutes.
- Round every published figure to one decimal place, half away from zero.
- carrier_plus_late_aircraft_delay_share_pct sums the carrier and late-aircraft minutes before dividing, rather than adding the two rounded shares.
What each column means
Two definitions decide whether a reuse is correct: cause shares divide by filed delay minutes rather than by all delay minutes, and 'weather' means extreme weather only.
| Column | Type | Definition |
|---|---|---|
| carrier_iata | string | IATA carrier code as filed with BTS. |
| carrier_name | string | Marketing name of the reporting carrier. |
| flights | integer | Scheduled domestic flights the carrier filed in the reporting period. The denominator for every rate in this file. |
| arrivals_late_over_15_min | integer | Flights arriving more than 15 minutes after scheduled arrival. Cancelled and diverted flights are not counted as late. |
| late_pct | decimal | arrivals_late_over_15_min / flights, as a percentage to one decimal place. |
| cancellations | integer | Flights cancelled, as filed. |
| cancelled_pct | decimal | cancellations / flights, as a percentage to one decimal place. |
| diversions | integer | Flights diverted to an airport other than the scheduled destination. |
| total_arrival_delay_minutes | integer | Sum of arrival delay in minutes across all flights. Early arrivals count as zero rather than as negative minutes, which would flatter the mean. |
| mean_arrival_delay_minutes | decimal | total_arrival_delay_minutes / flights. Includes on-time flights as zero, so this is not the mean delay of delayed flights. |
| reported_delay_minutes | integer | Sum of the five filed delay causes. Smaller than total_arrival_delay_minutes, because a cause is only filed for flights delayed 15 minutes or more. This is the denominator for every *_share_pct column. |
| carrier_delay_minutes | integer | Delay minutes coded to circumstances within the carrier's control — crew, maintenance, fuelling, baggage loading. |
| late_aircraft_delay_minutes | integer | Delay minutes coded to the inbound aircraft arriving late from a previous leg. |
| nas_delay_minutes | integer | Delay minutes coded to the National Airspace System: air-traffic volume, airport operations, and non-extreme weather handled through traffic management. |
| weather_delay_minutes | integer | Delay minutes coded to extreme weather only. Weather that is managed through air-traffic flow control is filed under NAS, so this column understates all weather-driven delay. |
| security_delay_minutes | integer | Delay minutes coded to security screening, terminal evacuation or a breach. |
| carrier_delay_share_pct | decimal | carrier_delay_minutes / reported_delay_minutes, as a percentage. |
| late_aircraft_delay_share_pct | decimal | late_aircraft_delay_minutes / reported_delay_minutes, as a percentage. |
| nas_delay_share_pct | decimal | nas_delay_minutes / reported_delay_minutes, as a percentage. |
| weather_delay_share_pct | decimal | weather_delay_minutes / reported_delay_minutes, as a percentage. |
| security_delay_share_pct | decimal | security_delay_minutes / reported_delay_minutes, as a percentage. |
| carrier_plus_late_aircraft_delay_share_pct | decimal | The two filed categories carrier_delay_minutes and late_aircraft_delay_minutes, summed and divided by reported_delay_minutes. THIS SUM IS OURS, NOT A BTS FIELD. It establishes no cause and no fault: late-arriving aircraft records that the inbound aircraft was late, not what delayed it, so an unknown part of it began as weather or traffic flow earlier in the day. Both components are published separately so the sum can be rejected. |
What this data cannot tell you
- Domestic US scheduled service only. 14 CFR part 234 requires reporting from US carriers above 0.5% of domestic scheduled-service passenger revenue, and covers domestic service. No international flight appears in this file.
- It therefore says nothing about any US-Mexico route, and nothing at all about Volaris, Viva Aerobus or Aeroméxico. Mexican carrier performance comes from AFAC, a separate source not published here.
- One reporting month, carrying that month's seasonality. A snapshot, not a trend.
- Only carriers required to report appear. A hub served largely by non-reporting regional carriers will read better here than a traveller experiences.
- Delay-cause shares exclude flights delayed under 15 minutes, for which no cause is filed.
- 'Weather' means extreme weather as BTS defines it. Weather managed through air-traffic flow control is filed under NAS, so weather's true contribution is higher than the weather column alone.
- carrier_plus_late_aircraft_delay_share_pct sums two BTS fields; the sum is ours, not a federal statistic, and it establishes no fault.
Cite this dataset
Naming the derived index rather than BTS alone, because the grouping of late-arriving aircraft with carrier delay is ours and a reader should be able to find out what it means.
Travelaidnow (2026). Airline Accountability Index — US domestic on-time performance by cause, May 2026 (v1.0.0). Derived from US DOT Bureau of Transportation Statistics, Reporting Carrier On-Time Performance. Retrieved 2026-07-28. https://www.travelaidnow.com/data/airline-accountability
May 2026 · CC0 1.0 — This derived aggregation is dedicated to the public domain under CC0 1.0: copy, modify and redistribute it, commercially or not, without permission or attribution. Citation is requested as a courtesy, not required. The dedication covers this data only — not the prose, code or design of travelaidnow.com, and not the Travelaidnow name.
The underlying BTS filings are works of the US Government and carry no copyright under 17 U.S.C. §105. That is their status, not a licence we grant.
Embed this chart
Paste this into any article or post. It renders the cause breakdown above, updates when the reporting period does, and carries the source and CC0 terms with it. No script, no tracking, no sign-up.
<iframe src="https://www.travelaidnow.com/embed/airline-accountability" title="Airline Accountability Index — May 2026" width="100%" height="620" loading="lazy" style="border:0"></iframe>
<p><a href="https://www.travelaidnow.com/data/airline-accountability" rel="nofollow">Airline Accountability Index</a> — US DOT BTS filings, compiled by Travelaidnow. CC0 1.0.</p>Keep the second line. A link inside a frame belongs to the frame, so the paragraph beneath it is what credits the source and lets your readers reach the limitations behind the figures. It is marked nofollow, so pasting it passes no ranking signal to us and asks nothing of your SEO.
Free to reproduce, modify and republish, commercially or not, without asking. May 2026 figures.
Delayed on one of these airlines?
Whether the cause was the airline's own matters legally, not just reputationally: Mexican law owes compensation only where the delay is attributable to the carrier, and US law owes a refund past a threshold that differs for domestic and international trips. Our checker works out which applies and cites the provision.
These are monthly aggregate filing codes. They cannot establish what caused your individual flight, whether it was within the carrier's control, or whether compensation is owed.
Check what you are owedFAQ
Which airline is worst? — the questions people ask
Answered from the figures on this page rather than from reputation, so the answers move when the federal data does.
Which US airline has the most delays?
Southwest Airlines, on the latest federal filing: 27.5% of its departures left more than fifteen minutes late, against 16.7% for Alaska Airlines at the other end of the table. That is a reported figure rather than a survey of opinion, covering 120,646 flights in May 2026. Worth knowing before you read a "worst airline" list elsewhere: most are built from complaint counts or social media sentiment, which track how many people fly a carrier as much as how well it runs.
Which airline cancels the most flights?
American Airlines, at 1.5% of departures, with JetBlue Airways lowest at 0.2%. Notice that this is a different question from which airline runs late, and usually a different answer — Southwest Airlines is latest but not the most cancellation-prone. A delay costs you hours; a cancellation costs you the day and triggers a refund right the delay may not. If your itinerary has no slack, the cancellation column is the one to read.
Who has more delays, United or Delta?
United Airlines, by 2.4 percentage points: 19.9% of United departures ran late against 17.5% of Delta's. Cancellations run the other way, 0.4% for United against 1.2% for Delta, so neither carrier is straightforwardly better than the other and which one suits you depends on whether a lost hour or a lost day would hurt more.
Which airline should I avoid?
We will not answer that as a recommendation, because the figures do not support one. A spread of 16.7% to 27.5% late means the worst carrier here still leaves roughly seven flights in ten on time, and your route matters more than your airline — the same carrier is punctual at one hub and poor at another. What the data does support is checking your specific route before you book, which is what the route records are for, and knowing what the carrier has committed to owe you when it does go wrong.
Is a delay the airline's fault or just the weather?
The filings do not settle fault, but they do show that very little of it is filed as weather. Across every carrier on this page, far more delay minutes are coded to the carrier's own operation — crew, maintenance, fuelling — and to aircraft arriving late from a previous leg than to extreme weather. Read the second of those carefully: it records that the inbound aircraft was late, not what delayed it, so part of it may have begun as weather hours earlier. What none of it settles is any particular flight. Compensation under Mexican art. 47 Bis and European Regulation 261 turns on whether the cause was within the carrier's control, and that is decided on the facts of one flight — a month of aggregate filings cannot establish it, and no figure on this page should be read as evidence about a trip you took.