Highway Traffic Volume Aadt Data: Dataset Structure and Field Coverage

Five named sources answer the question, and only one of them publishes true segment-level AADT.

Highway Performance Monitoring System (HPMS) is the anchor. FHWA describes it as the official national inventory of all US public roads, carrying state-reported route-level AADT, through-lane counts, pavement IRI, functional system and ownership for all 50 states, DC and Puerto Rico. States have submitted annually since 1978, each record is linear-referenced between begin and end points, and values roll up cleanly to county, urbanized-area, state and national totals.

FHWA Highway Statistics Series (1945-present) supplies the official state-level context: vehicle-miles-of-travel (VM) tables among roughly 153 downloadable files per recent edition, with within-edition trend lines reaching back to 1920 for road mileage and 1957 for federal-aid VMT. It is the only perfect quality-score-10 record in this industry's pool.

The USDOT Open Data Portal republishes HPMS (dataset id 4dez-3n4e, category "Roadways and Bridges", annual accrual) inside a Socrata catalog of about 176 highway- and rail-related tables drawn from roughly 1,515 dataset records, queryable without an integration key.

Two further sources matter at the edges. The National Bridge Inventory (NBI) attaches traffic exposure to about 624,000 bridge records in its 2025 edition, and the OpenStreetMap family - Geofabrik extracts and the Overture Maps Transportation Theme - contributes the geometry you join counts onto rather than counts themselves: Overture's August 2026 class counts alone list about 128 million residential and 34.9 million service-road segments.

Each segment record pairs its traffic value with the descriptors a demand model or pavement model actually needs:

• AADT - Field Manual Item 21, annual average daily traffic for all vehicles.
• AADT_SINGLE_UNIT and AADT_COMBINATION - Items 22 and 24, splitting truck exposure into single-unit/bus flows and combination-truck flows.
• THROUGH_LANES (Item 7), F_SYSTEM (Item 1), OWNERSHIP (Item 6), NHS and TRUCK designations, plus URBAN_CODE, COUNTY_CODE, ACCESS_CONTROL, SPEED_LIMIT and the TOLL_CHARGED/TOLL_TYPE pair.
• Pavement companions: IRI (Item 47) with its IRI_YEAR, and PSR (Item 48).
• ROUTE_ID with BEGIN_POINT/END_POINT, tying every value to the linear referencing system.

Observed rows from the Alabama 2018 public release show the spread the fields capture: an Interstate segment (IN0000650000) carries AADT 93,456 across six through lanes with an IRI of 46, while a rural two-lane segment (AL0001830000) records AADT 398 with an IRI of 119 - more than double the Interstate's roughness reading, moving in the opposite direction from volume.

Coverage is deepest on the higher systems. Full-extent attributes cover the Federal-Aid System and NHS roads, arterials and collectors enter through point-level sample panels, and lower functional systems carry summary coverage only - worth knowing before you sum the file and call the result national VMT.

Demand modeling, pavement planning, freight screening and international benchmarking each pull from a different shelf of this pool. The comparison below lines up the six sources that touch traffic volume directly, with grain, measures, history and terms side by side.

Demand modeling and pavement planning stay with HPMS: AADT by functional system is the exposure term in both, and the IRI and PSR fields let one table drive volume-and-condition joins without a second source. Freight screening starts at the USDOT portal, where 438,762 grade-crossing rows (187 columns) meet 251,149 Form 57 incident rows (154 columns) through one API, then moves to NPMRDS for delay on the corridors themselves. Multi-decade trend work needs Highway Statistics, because HPMS submissions only run back to 1978 while the state tables reach 1945. European comparisons have no route-level equivalent in this pool - Eurostat's bulk tables are country-by-year aggregates - so benchmark capacity, not counts.

Three questions recur whenever analysts first load HPMS - granularity, per-route extraction, and how the sums reconcile against published VMT. Each has a concrete answer grounded in the cataloged sources above rather than in vendor claims.

Start with HPMS for segment AADT, add Highway Statistics when you need eighty years of state context, and reach for NPMRDS only when monthly truck performance justifies the account setup. Every primary source named here is free, inside a 15-dataset industry pool where pricing is uniformly zero. The wider landscape - including the OSM-based geometry layers to join counts against and the UK rail feeds on the other side of the industry - is mapped in the highways railtracks data guide.