Passenger Airlines Data Guide: Every Dataset, Source and Cost in 2026

Passenger airline data clusters into three families — government operational statistics, machine-learning mirror sets, and commercial travel APIs — and it is unusually cheap. Of the 14 datasets Datadory catalogs for this industry (6 primary records plus 8 related ones), five of the six primary datasets are free; only Amadeus charges beyond its quota. The record-by-record view lives in our passenger-airlines data hub.

The U.S. Bureau of Transportation Statistics anchors the official side. Its TranStats portal indexes 21 aviation databases, led by monthly Airline On-Time Performance records for certificated U.S. carriers back to October 1987 — roughly 5–7 million flight rows per year — and census-level T-100 traffic with load factors from 1990, all commercial delivery terms and downloadable without registration.

Reference and macro layers round out the stack. OpenFlights ships commercial delivery terms CSV files listing 7,698 airports, 6,162 airlines and 67,663 routes frozen at June 2014, while the World Bank indicator IS.AIR.PSGR tracks registered-carrier passengers across roughly 265 economies from 1970 through 2023 under commercial delivery terms-4.0.

How much of this comes from outside the industry slice?

Eight of the 14 records are cross-listed from neighbouring industries, which matters for coverage planning. EUROCONTROL Aviation Data & Dashboard, FAA Air Traffic By The Numbers, FAA Aircraft Registry, Boeing Commercial Market Outlook and Airbus Global Market Forecast all sit under aerospace-defense; UK CAA Aviation Data & Analysis sits under air-freight-logistics; FlightAware Live Flight Tracking also arrives via air-freight-logistics; and AviationStack Flight & Airport API is cross-listed from agricultural-products-services. Several primary records carry their own secondary industries too — BTS TranStats maps to aerospace-defense and airport-services, while both Kaggle mirrors map to airport-services.

Government operational statistics

EUROCONTROL Aviation Data & Dashboard adds the European complement at quality score 9: daily dashboards plus bulk XLSX, CSV and Parquet series with en-route delays from January 2011 and more than 25 million flights of trajectory data in the ADRR.

UK CAA Aviation Data & Analysis is a free open-government portal publishing monthly and annual CSV/PDF tables of passengers and air transport movements for roughly 50–60 UK airports, archived to 1973.

Commercial and API-native feeds

Public benchmarks and reference files

2015 Flight Delays and Cancellations (5.8M U.S. Flights), quality score 9 under commercial delivery terms, packages calendar-2015 domestic flying as three files around 200 MB zipped: flights.csv with 5,819,079 rows (~586 MB uncompressed), a 14-row airlines.csv and a 322-row airports.csv with coordinates. The snapshot was published February 2017 and never changes.

OpenFlights Airline, Airport and Route Database, quality score 8 under the commercial delivery terms, distributes 7,698 airports (~700 KB), 6,162 airlines (~400 KB) and 67,663 directional routes (~2 MB) with IATA/ICAO codes, coordinates and codeshare flags. Airport and airline files get periodic refreshes (the docs cite a January 2017 airport baseline); the route table comes from a third-party feed frozen in June 2014.

World Bank — Air Transport, Passengers Carried (IS.AIR.PSGR) serves roughly 265 economies × 54 years, about 14,000 observations, as JSON, XML, CSV or Excel with no integration key, under commercial delivery terms-4.0. The WDI database last refreshed on 2026-07-13.

Manufacturer forecasts close the set: Boeing Commercial Market Outlook projects 2026–2045 across eleven regions and 50+ origin-destination flows as PDF plus Excel after 65+ years of publication, and Airbus Global Market Forecast covers the same horizon with regional delivery projections, earlier editions such as 2023–2042 still archived.

Ten use cases recur across this slice, each tied to a named record:

1. Training flight delay and cancellation classifiers on the Kaggle/USDOT 2015 dataset's 5,819,079 per-flight rows, then extending back through the BTS on-time archive to October 1987 for deeper history.
2. Benchmarking certificated-carrier capacity, segment traffic and load factors with BTS T-100 tables covering passengers, seats and ASK from 1990 onward.
3. Building satisfaction-prediction models on the Airline Passenger Satisfaction survey's 129,880 responses across 23 service attributes, using its built-in 103,904/25,976 train/test split.
4. Sizing and comparing national air travel markets with the World Bank's IS.AIR.PSGR passengers-carried series for roughly 265 economies from 1970 to 2023.
5. Mapping network reach, equipment types and codeshare structure using OpenFlights' 67,663 directional routes across 7,698 airports and 6,162 airlines.
6. Prototyping fare-shopping, flight-status and seatmap features against Amadeus's free test-environment quotas (10 TPS) before committing to production billing at 40 TPS.
7. Stress-testing 20-year traffic and fleet assumptions against Boeing's Commercial Market Outlook and Airbus's Global Market Forecast Excel workbooks covering 2026–2045.
8. Monitoring European schedule reliability and emissions with EUROCONTROL's daily dashboards and bulk en-route delay series starting January 2011.
9. Geocoding origin-destination studies by joining the 2015 delays set's airports.csv coordinates (322 airports) or OpenFlights' 7,698-airport reference file.
10. Tracing UK airport passenger volumes and air transport movements with UK CAA monthly releases reaching back to 1973.

Market researchers and analysts ask different questions: "airline passengers carried by country" resolves to the World Bank's IS.AIR.PSGR series for roughly 265 economies, "t100 airport pair load factor data" points at BTS segment tables, and "commercial aviation demand forecast data" leads to the Boeing and Airbus 2026–2045 Excel workbooks. Our page for market researchers collects those threads.

A third group — network planners and aviation economists — works the same records from the capacity side, combining T-100 load factors with OpenFlights' 67,663 directional routes and EUROCONTROL's European en-route delay series from January 2011.

What do the long-tail searches reveal?

A workable stack assembles one layer at a time, cheapest first:

1. Reference backbone. Load OpenFlights' three CSVs — 7,698 airports, 6,162 airlines, 67,663 routes — for codes, coordinates and codeshare flags; remember the route file froze in June 2014.
2. Official operating history. Pull BTS TranStats extracts: per-flight on-time rows from October 1987 (~5–7 million per year) and census-level T-100 segments with load factors from 1990, monthly-refreshed and commercial delivery terms.
3. Model-ready benchmarks. Add the commercial delivery terms 2015 Flight Delays set (flights.csv ~586 MB uncompressed plus 14-row airline and 322-row airport lookup files) for reproducible classifier baselines, and the 129,880-row satisfaction survey with its built-in 103,904/25,976 split for customer-experience models.
4. Macro layer. Join the World Bank IS.AIR.PSGR panel (~265 economies, 1970–2023, ~14,000 observations, keyless JSON/XML/CSV/Excel) for cross-country demand framing.
5. Real-time prototype. Build fare-shopping, status and seatmap features against Amadeus's free test quota at 10 TPS, then move to production at 40 TPS once volumes justify pay-as-you-go billing.
6. Extend and plan. Layer EUROCONTROL's bulk XLSX/CSV/Parquet delay series (from January 2011) and UK CAA tables (to 1973) for Europe, and stress-test assumptions against Boeing CMO and Airbus GMF workbooks spanning 2026–2045.

Which joins actually work between these sources?

Three joins carry most projects. Airport-level geocoding: the 2015 delays set ships its own 322-row airports.csv with coordinates, which merges cleanly against OpenFlights' 7,698-airport file on IATA codes. Carrier identification: airlines.csv in the Kaggle mirror lists 14 reporting carriers, expandable through OpenFlights' 6,162-airline file for ICAO codes and callsigns. Demand framing: T-100 segment load factors roll up to country comparisons that the World Bank panel (~14,000 observations) validates from 1990 onward.

Every figure below comes straight from the cataloged records; the denominators are the ones Datadory tracks.

Quoted as declarative sentences: BTS TranStats indexes 21 aviation databases and accumulates roughly 5–7 million on-time flight rows per year. The 2015 Flight Delays mirror holds exactly 5,819,079 rows under commercial delivery terms. Amadeus caps its test environment at 10 transactions per second and bills production at 40 TPS past a free monthly quota.

Continue through the rest of the guide set: