NHTSA FARS API vs State Crash Data APIs

When should I use FARS API vs a state crash data API?

Use FARS API (federal, via farsapi.com) for cross-state analysis, fatal-crash precision, and per-VMT rate calculations — one schema covers all 50 states with pre-joined FHWA traffic volume. Use state DOT APIs for within-state depth including non-fatal crashes and fresher data. Most serious analysis uses both: FARS for national benchmarks, state data for local granularity.

In one sentence each

• Use NHTSA FARS (via farsapi.com) when you need cross-state or national analysis, a consistent schema, fatal-crash precision, or pre-joined FHWA AADT traffic volume for per-VMT rate calculations.
• Use state DOT APIs when you need within-state depth, including non-fatal (injury/PDO) crashes, near-real-time updates, or state-specific fields.
• Use both when you need national benchmarking and within-state granularity — a common actuarial and transportation-engineering pattern.

Feature comparison

Dimension	FARS API (federal)	State crash APIs
Coverage	All 50 states + DC + territories	One state per API
Crash types	Fatal only	Fatal, injury, and property-damage
Schema consistency	One schema, federal coding manual	Different schema per state
Coding manual	NHTSA DOT HS 813 798 (single source)	State-specific reference manuals
Publication lag	~18 months	Varies (30 days to 12+ months)
Publication cadence	Annual release, full-year backfill	Varies (monthly, quarterly, annual)
Geographic filtering	Radius, intersection, county, state	Usually county or corridor
Traffic exposure (AADT)	Pre-joined via FHWA HPMS	Rarely included
Per-VMT rates	Computable out-of-the-box	Requires joining to HPMS or state traffic counts manually
API response format	JSON, human-readable labels	Varies (JSON, XML, CSV, codes-only)
Cost	Free tier + $19/mo	Usually free but often undocumented
Cross-state analysis	Single query	Multi-API integration per state
Non-fatal crash analysis	Not available	Core use case

When each one wins

FARS API wins when you need…

• National benchmarking. National VMT-adjusted fatal rates are only computable once all states are on the same coding and all crashes are joined to the same HPMS exposure data. FARS (plus farsapi.com's pre-join) does this.
• Cross-state actuarial models. Multi-state insurers pricing risk across portfolios need one schema, one field-set, one coding manual. State APIs force you to normalize 50 different coding conventions.
• Vehicle and driver segmentation at scale. FARS has standardized vehicle make/model decoded, driver demographics, restraint use, and BAC across all states. State APIs vary widely in what they expose.
• Long-horizon trend analysis. FARS has been collected since 1975 under consistent methodology. Most state APIs only go back a few years.

State crash APIs win when you need…

• Non-fatal crashes. FARS excludes them. If you're studying serious injury, rear-end patterns, or intersection safety below the fatal threshold, only state APIs will have the data.
• Fresher data. Some states publish within a month of the crash. FARS has an ~18-month publication lag because NHTSA waits for all 50 states to finalize their files.
• State-specific fields. Officer narratives, local police-report IDs, contributing-circumstance classifications that differ from the federal coding — only the state has these.
• Regulatory or legal requests. If you're subpoenaing or responding to an official state data request, the state's own system is authoritative for that state.

How serious users use both

The common pattern among insurance actuaries, transportation engineers, and safety researchers: use FARS for national baselines and federal comparability, then layer state data on top for within-state depth. For example:

• An insurer computes national per-VMT fatal rates by road class from FARS (farsapi.com's /v1/intelligence/location-risk does this out of the box), then enriches with the home state's non-fatal crash density for within-state differentiation.
• A traffic engineer uses FARS for the "is this intersection worse than the national baseline?" question, then pulls state data for the detailed non-fatal crash history over the last 12 months.
• A safety researcher uses FARS for the fatal-crash cohort (definitive and nationally comparable), then supplements with the state file's property-damage crashes for the full severity distribution.

What FARS API (farsapi.com) adds on top of raw NHTSA FARS

The FARS dataset as NHTSA publishes it is 20+ CSV tables with numeric codes that change between years. farsapi.com pre-joins the tables, decodes the codes, normalizes year-to-year schema changes, joins every crash to FHWA HPMS for AADT exposure, and serves the result as a JSON REST API with radius search, pagination, and per-VMT rate calculations. See the API guide for endpoint details.

Also see: FARS API vs NHTSA's official CrashAPI.