Open traffic and mobility datasets
For transit schedules, the Mobility Database is the current registry of real GTFS feeds worldwide; for ride-level data, NYC TLC's taxi and for-hire-vehicle records are the largest public trip archive; for raw GPS trajectories, Microsoft's T-Drive dataset remains the standard research benchmark; and for road-sensor traffic flow, Caltrans PeMS is the reference source in the US. Below is a source-by-source look at what each one holds and what it takes to access it.
The short answer
Match the source to the data type: transit schedules and real-time feeds (Mobility Database / GTFS), taxi and ride trip records (NYC TLC), raw GPS trajectories (T-Drive), road-sensor flow and speed (Caltrans PeMS), and micromobility trip data (Citi Bike, Divvy, and other GBFS-compliant bike/scooter share systems). All are free; a few require a one-time account signup and none require payment.
Sources, one by one
1. Mobility Database — GTFS transit feeds worldwide
The Mobility Database, maintained by the nonprofit MobilityData, is the current open catalog of General Transit Feed Specification (GTFS) data — schedules, stops, routes and fares — plus GTFS-Realtime and GBFS (bike/scooter share) feeds, covering more than 6,000 feeds across 99+ countries. It succeeded the older TransitFeeds registry, which stopped updating in 2024. Each entry links to the transit agency's own feed rather than hosting a copy that can go stale, and the catalog itself is free to browse and query with no account required. GTFS itself, described at gtfs.org, is used by over 10,000 agencies, which makes it the most consistent way to compare service coverage or build a multi-agency trip planner.
2. NYC Taxi and Limousine Commission (TLC) trip records
The TLC Trip Record Data is one of the largest public ride-trip archives available: monthly yellow-taxi, green-taxi and for-hire-vehicle (Uber/Lyft-class) records going back to 2009, with pickup/drop-off times, zone-level locations, distances, fares and payment types. Since May 2022 the files are published in Apache Parquet rather than CSV. It is published directly by New York City government as open data, free with no account required, and a full mirror is also available through the AWS Open Data Registry for direct S3 access (an AWS account is needed there, but the data itself remains free).
3. T-Drive GPS trajectory dataset (Microsoft Research)
The T-Drive trajectory dataset, released by Microsoft Research Asia, is a one-week public sample of GPS traces from 10,357 taxis in Beijing — about 15 million points covering roughly 9 million kilometers of driving, recorded February 2-8, 2008. Each point carries a taxi ID, timestamp, longitude and latitude, with no further personal information. It remains the most widely cited public GPS trajectory dataset for route-recommendation, travel-time-estimation and trajectory-mining research, precisely because it is dense, long enough to show real driving patterns, and released with no access barrier — download directly from the Microsoft Research page, free for research use.
4. Caltrans PeMS — freeway sensor network
The Performance Measurement System (PeMS), run by Caltrans, collects continuous flow, speed and occupancy data from nearly 40,000 loop detectors across California's freeway network, going back to the early 2000s. It is the reference dataset for road-sensor-based traffic-flow forecasting research, since it is one of the few sources offering years of dense, consistently formatted sensor data at this scale. Access requires a free account registration at the PeMS Data Clearinghouse; there is no charge for the data itself. Curated PeMS subsets (PeMS-BAY, PeMSD4/PeMSD7/PeMSD8) also circulate in the traffic-forecasting research community as pre-processed benchmark files.
5. Citi Bike and Divvy system data (bike share, GBFS)
Citi Bike System Data (New York) and Divvy Trip Data (Chicago) both publish full historical trip-level records — start/end station, timestamps, ride duration and rider type — as downloadable CSVs, with staff trips, test-station trips and sub-minute trips already filtered out. Both systems also expose live station and bike availability through the open GBFS (General Bikeshare Feed Specification) standard, so you can pair historical trip patterns with real-time capacity data. Both are free with no account required; Divvy's historical files are also cataloged on the City of Chicago's open data portal.
Sources side by side
| Source | Best for | Coverage | Access |
|---|---|---|---|
| Mobility Database (GTFS) | Transit schedules, routes, real-time feeds | 6,000+ feeds, 99+ countries | Free, no account |
| NYC TLC trip records | Taxi and for-hire-vehicle trip data | NYC, monthly files since 2009 | Free, no account (AWS account for S3 mirror) |
| T-Drive (Microsoft Research) | Raw GPS trajectory research | 10,357 taxis, 1 week, Beijing | Free, research use |
| Caltrans PeMS | Road-sensor flow, speed, occupancy | ~40,000 detectors, California | Free account required |
| Citi Bike / Divvy | Micromobility trip and station data | NYC / Chicago, full history | Free, no account |
How to choose
Start with what kind of movement you are studying. Scheduled public transit — buses, subways, rail — belongs in GTFS territory through the Mobility Database, which is the only source here that spans thousands of cities rather than one. On-demand ride data (taxis, ride-hail) is best represented by NYC TLC, simply because no other public archive matches its combination of scale and history; if your study area is somewhere other than New York, check whether the local transit or transportation authority publishes an equivalent open dataset before assuming none exists. If you need raw point-level GPS traces rather than aggregated trip records — for route-mining or trajectory-prediction research — T-Drive is the standard starting benchmark, though it is limited to one week and one city, so treat it as a methods testbed rather than a source of current conditions. Road-sensor-based congestion and flow forecasting should use PeMS, and any micromobility question belongs with the GBFS-based bike-share systems.
Before committing to any of these, verify three things: the geographic and time coverage actually matches your question (T-Drive, for instance, is 2008 Beijing only), whether the license permits your intended use (most of the above are open government or research-use data, but confirm on the source page rather than assuming), and whether you need real-time or historical data — several of these sources (GTFS-Realtime, GBFS, PeMS) offer live feeds in addition to historical archives, and the access pattern for each differs.
Frequently asked questions
Is Uber Movement still a usable source?
No. Uber Movement, which published city-level travel-time and speed data, has been discontinued and is no longer accepting new downloads. For similar city-scale mobility patterns, use NYC TLC trip records, GTFS transit feeds through the Mobility Database, or GPS trajectory datasets such as T-Drive instead.
What is GTFS and where do I get real feeds?
GTFS (General Transit Feed Specification) is the open standard transit agencies use to publish schedules, stops and routes. The Mobility Database, which succeeded the older TransitFeeds registry, catalogs more than 6,000 GTFS, GTFS-Realtime and bike/scooter-share (GBFS) feeds across 99+ countries, each linking back to the issuing agency's own feed.
Do I need a license to use NYC taxi trip data?
No special license or account is required. NYC TLC publishes trip records directly on nyc.gov as open government data in Parquet format, and the same files are mirrored on the AWS Open Data Registry for direct access, also free, though S3 access there requires an AWS account.
Can't find coverage for your city or route network? Leave it to us
If your project needs mobility data for a city, region or mode that is not covered here, you can give us your task goal and the conditions it must meet, and we run a data availability assessment (3 free credits on sign-up) first — searching reputable data platforms for real and judging matches and gaps against each required item. Even when no perfectly fitting dataset is found, we present the search directions, approximate sources and item-by-item findings honestly for your reference.
New accounts get 3 free searches; every result comes with verifiable source links.