NHTSA Vehicle Safety API (Recalls, Complaints, Ratings)

Datadory delivers automotive-parts-and-equipment data covering US vehicle safety recall campaigns from model year 1949 onward, ODI owner complaints with crash, fire, injury, and fatality flags, and tire-level product records. Every campaign number, defect narrative, and complaint arrives cleaned and keyed to make, model, and model year.

What is the NHTSA vehicle safety dataset?

It is the United States government's complete public record of motor-vehicle safety defects, structured three ways. Recall campaigns document each officially announced defect: what broke, who must fix it, how, and whether owners are told to park the vehicle. Owner complaints are the raw grievance stream behind those campaigns — millions of filings in the Office of Defects Investigation archive, each with its own incident date, component list, and narrative. Vehicle listings tie both streams to the full make-model-model-year grid, 80 distinct model years deep.

The detail level surprises people the first time. A recall record does not just say "fuel pump recalled" — it names the component in NHTSA's taxonomy (FUEL SYSTEM, GASOLINE:DELIVERY:FUEL PUMP), quotes the manufacturer's own consequence language, describes the remedy the dealer will perform, and flags whether the fix ships over the air. A complaint record does not just say "car caught fire" — it carries boolean crash and fire markers, counted injuries and deaths, a truncated VIN, and a FOIA-redacted consumer story.

Because every complaint distinguishes the vehicle from any tire or equipment product involved, the dataset reaches past OEMs. One 2019 Corolla complaint we sampled lists both the Toyota itself and a Yokohama AVID ASCEND LX in size 205/55 R16 — the tire appears as a first-class record, not prose.

Sample rows

Four real records, one from each record family. First a recall campaign, then two ODI complaints — note the second one carrying the tire product array — then a vehicle-listing row:

NHTSACampaignNumber   20V024000
Manufacturer          Toyota Motor Engineering & Manufacturing
Component             AIR BAGS: AIR BAG/RESTRAINT CONTROL MODULE
ReportReceivedDate    17/01/2020
Consequence           may increase the risk of injury
Remedy                dealers will inspect the ECU and install a noise filter
parkOutSide           false · overTheAirUpdate false

odiNumber             11754726
dateComplaintFiled    08/04/2026
components            ELECTRICAL SYSTEM
crash false · fire false · numberOfInjuries 0 · numberOfDeaths 0
vin                   5YFBURHE4KP

odiNumber             11551548
components            TIRES
products[0]           Vehicle · TOYOTA COROLLA · year 2019
products[1]           Tire · YOKOHAMA AVID ASCEND LX · 205/55 R16

catalog row           modelYear 2019 · TOYOTA · COROLLA

What fields does the dataset include?

Twenty-nine documented fields across four record families: recall campaigns, ODI complaints, the product arrays nested inside complaints, and the vehicle catalog listings. The dictionary below shows the core of it — the fields that carry most of the analytical weight.

How is the data covered?

Geography: United States vehicles and motor-vehicle equipment, including tires, sold or operated in the US. Temporal: recall campaigns span model years 1949 through 2027 — seven and a half decades of safety history — plus a pseudo-year bucket (9999) holding tire and equipment records untied to any vehicle year. Complaint filings run through August 2026 at time of research. Granularity: one row per event, never aggregated — each recall campaign, each individual complaint, each make-model-year listing is its own record you can count, filter, and join.

Scale check: a single query on one popular model returned 5 recall campaigns against 201 owner complaints. That 40-to-1 ratio between grievances and official action is exactly why teams join the complaint stream against their own warranty and claim data — the complaints arrive years ahead of the campaigns.

How is the data delivered?

API, files, or your warehouse. Daily, weekly, or hourly.

Who uses this data?

  • Aftermarket parts and fitment teams watch the Component taxonomy across new campaigns so a part they stock or manufacture never surprises them in a recall notice. The equipment entries (EQUIPMENT:OTHER:LABELS) reach suppliers with no vehicle of their own.
  • Warranty and quality analysts pull complaint clusters by component and model year, then match them against internal claim files. Crash, fire, injury, and death flags let severity weighting happen at query time instead of by reading narratives.
  • Competitive intelligence teams track rival manufacturers' campaign cadence — how often a competitor recalls, which systems fail, whether fixes ship over the air or require a dealer visit. Campaign numbers sequence the timeline automatically.
  • Tire and rubber industry analysts get something rare: tire brands appearing as named records (YOKOHAMA, CONTINENTAL) with model lines and sizes, turning consumer gripes into supplier-level signal.
  • Journalists and academics cite the per-complaint injury and fatality counts directly — they are reported values, not estimates — and the filing-versus-incident dates expose reporting lag.

What makes this dataset hard to work with raw?

Three things, all solved by taking it through Datadory instead. Dates arrive as DD/MM/YYYY strings mixed into otherwise JSON-shaped responses, so naive parsers silently swap days and months — every date here arrives typed. Field presence is inconsistent: some complaint records lack VINs, some recall records lack action numbers, and the two families use different capitalization conventions for the same concepts (Component versus components). We normalize both families into one schema. The ratings surface is undocumented — the listing endpoints answer cleanly but several rating paths respond with access errors, so rating retrieval has to be learned behaviorally rather than from a reference; our pipeline handles whichever paths resolve, and folds the rest under additional fields on request.

Sample rows — recall campaign, ODI complaints (one with tire products), vehicle listing

Record familyKey fieldsValues
Recall campaignNHTSACampaignNumber20V024000
Recall campaignComponentAIR BAGS: AIR BAG/RESTRAINT CONTROL MODULE
Recall campaignRemedydealers will inspect the ECU and install a noise filter
ODI complaintodiNumber / filed11754726 / 08/04/2026
ODI complaintflagscrash=false fire=false injuries=0 deaths=0
ODI complaint + tireodiNumber11551548
ODI complaint + tireproducts[]Vehicle: TOYOTA COROLLA 2019; Tire: YOKOHAMA AVID ASCEND LX 205/55 R16
Vehicle listingmodelYear / make / model2019 / TOYOTA / COROLLA

Field dictionary — core fields (full dictionary with examples delivered with your sample)

FieldTypeDefinitionExample
NHTSACampaignNumberstringRecall record: NHTSA campaign identifier for the recall.20V024000
NHTSAActionNumberstringRecall record: NHTSA investigation/action number; observed present on some records only.EA19001
ManufacturerstringRecall record: manufacturer of record that filed the campaign (may be a distributor such as Southeast Toyota Distributors, LLC).Toyota Motor Engineering & Manufacturing
ReportReceivedDatedateRecall record: date NHTSA received the recall report (DD/MM/YYYY string).17/01/2020
ComponentstringRecall record: affected component or system, using NHTSA component taxonomy (covers vehicle systems and equipment such as EQUIPMENT:OTHER:LABELS).FUEL SYSTEM, GASOLINE:DELIVERY:FUEL PUMP
SummarytextRecall record: description of the safety defect.
ConsequencetextRecall record: description of the injury or crash risk posed by the defect.may increase the risk of injury
RemedytextRecall record: description of the fix and owner contact instructions.dealers will inspect the ECU and install a noise filter
parkItbooleanRecall record: whether owners are advised to park the vehicle.false
parkOutSidebooleanRecall record: whether owners are advised to park outside (fire risk).false
overTheAirUpdatebooleanRecall record: whether the remedy can be delivered as an over-the-air software update.false
odiNumberintegerComplaint record: Office of Defects Investigation unique complaint identifier.11754726
crashbooleanComplaint record: whether a crash was reported.false
firebooleanComplaint record: whether a fire was reported.false
numberOfInjuriesintegerComplaint record: reported injury count.0
numberOfDeathsintegerComplaint record: reported fatality count.0
dateOfIncidentdateComplaint record: date of the incident (DD/MM/YYYY string).04/04/2023
dateComplaintFileddateComplaint record: date the complaint was filed with ODI (DD/MM/YYYY string).08/04/2026
vinstringComplaint record: truncated 11-character VIN of the complainant vehicle; null on some records.5YFBURHE4KP
componentsstringComplaint record: comma-separated list of affected NHTSA components.ELECTRICAL SYSTEM
summarytextComplaint record: consumer narrative; personal details redacted.
products[].typeenumComplaint record: product type of each involved product; observed values Vehicle and Tire.Tire
products[].productMakestringProduct record: make of the involved vehicle or tire brand (TOYOTA, YOKOHAMA, CONTINENTAL).YOKOHAMA
products[].productModelstringProduct record: model or tire line name (COROLLA, AVID ASCEND LX, CONTI PRO CONTACT).AVID ASCEND LX
products[].sizestringTire product record only: tire size designation.205/55 R16
modelYearintegerVehicle listing record: model year (listing records return integer; recall records return string).2019
vehicleModelstringVehicle listing record: model name in uppercase.COROLLA

Coverage chips

DimensionCoverage
GeographyUnited States — vehicles and motor-vehicle equipment, including tires, sold or operated in the US
TemporalModel years 1949–2027 for recall campaigns (plus a 9999 pseudo-year for equipment/tire records); complaint filings through August 2026 at time of research
GranularityOne row per event: each recall campaign, each ODI complaint, each make/model/year listing
Record families4 — recall campaigns, ODI complaints, nested product records, vehicle listings

Additional fields available on request

Field groupNotes
Count / Message envelopeResponse envelope fields (result count, status message) included in file deliveries on request.
Ratings surfacesRating retrieval paths are learned behaviorally rather than from documentation; whatever resolves is folded into delivery, remainder on request.
Extended product attributesAdditional product-array attributes beyond make/model/size (e.g. productYear, manufacturer-of-record) included with sample.

Questions buyers ask

What fields does the NHTSA vehicle safety dataset include?

Recall records carry campaign numbers, affected components, defect narratives, consequence statements, remedy descriptions, and park-it / park-outside / over-the-air-update advisories. Complaint records carry ODI numbers, crash and fire flags, injury and death counts, incident and filing dates, truncated VINs, affected components, and redacted consumer narratives. Product records name the involved vehicle or tire down to brand, model line, and size.

How far back does NHTSA recall data go?

Recall coverage spans model years 1949 through 2027, with an additional pseudo-year bucket holding equipment and tire records that are not tied to a specific vehicle year. A query on one recent model returned 5 recall campaigns against 201 owner complaints for the same vehicle, which gives a sense of the ratio between official campaigns and filed grievances.

Does the data cover tires and equipment, not just vehicles?

Yes. Complaint records distinguish Vehicle products from Tire products, naming the tire brand, model line, and size designation. A single complaint can list both — the involved vehicle and the exact tire fitted to it — making this one of the few regulatory datasets where aftermarket suppliers appear as named entities rather than footnotes.

Who uses NHTSA recall and complaint data?

Aftermarket parts teams use it to spot components under regulatory scrutiny before quoting fitment. Warranty and quality analysts track complaint clusters against their own claim files. Competitive intelligence teams monitor rivals' campaign cadence. Journalists and academics cite the injury and fatality counts, which are recorded per complaint rather than estimated.

Can I get this data joined to my own warranty or claims records?

That is the standard ask. Records key naturally on make, model, and model year, and complaint records add a truncated VIN for tighter joins. Tell us the shape of your claim file and we will return a matched sample showing exactly how many of your records pick up campaign and complaint history.

How current is the complaint stream?

Filings dated August 2026 appeared in the data at time of research, so the stream runs to within days of the present. Delivery cadence is your choice regardless — hourly if you are watching a live defect story break, weekly if you are feeding a quarterly quality review.

See the rows before you pay anything.

Name this dataset and we send real records from it — scoped to the fields you asked for.

See pricing