NHTSA Vehicle Complaints & Defect Investigations Feed (Early-Warning API)

Get the newest NHTSA vehicle safety complaints across every make and every NHTSA defect investigation (Preliminary Evaluations, Engineering Analyses, Recall Queries, Defect Petitions) as clean JSON, CSV or Excel. Complaints are the earliest public signal of a vehicle defect, often months before a recall, and investigations show where the regulator is already looking. Filter by make, model, model year, component (air bags, brakes, steering, fuel, ADAS...) and crash/fire/injury, and schedule it daily in new since last run mode for an alert feed that only bills you for records you have not seen.

Unlike make/model/year lookup tools, this Actor reads NHTSA's official daily ODI flat files, so it sees all ~6,000 complaints NHTSA receives each month, for all makes, without you knowing which vehicle to ask about. It also adds a complaint-cluster signal (similarComplaintsInWindow) so spikes stand out.

Who it is for

• Auto-defect, lemon-law and product-liability law firms: spot emerging defect clusters (e.g. 18 complaints about one model's fuel system in 30 days), crash/fire/injury complaints, and new investigations the day NHTSA opens them.
• OEM and supplier quality, warranty and safety teams: watch complaints about your vehicles, your competitors' vehicles or your component category; get notified when ODI opens a PE or upgrades to an EA.
• Insurers, claims and subrogation teams: fire and crash complaints by model and component, investigations tied to recall campaigns.
• Fleet managers and dealers: alerts for the makes/models/years you operate or sell.
• Journalists and researchers: a daily feed of new federal defect investigations and the complaints behind them, including robotaxi / ADS operator complaints (adsOperator).
• Developers: normalized NHTSA data through the Apify API, webhooks and integrations (Slack, Google Sheets, Zapier, Make).

What you get

• Consumer complaints (all makes): ODI number, date received, incident date, manufacturer, make, model, model year, product type (vehicle / tire / equipment / child seat), all components, crash, fire, injuries, deaths, medical attention, police report, towed, mileage, speed, incident state, complaint source and the narrative.
• Defect investigations: action number, type (PE, EA, RQ, DP, AQ...), status (open/closed), dates opened and closed, the event that put it in your window (opened or closed), manufacturer, subject, component, all affected makes/models/years, the resulting recall campaign (with link) and NHTSA's full summary.
• Complaint clusters: similarComplaintsInWindow = number of complaints in the same window for the same make, model, year and component.
• Full vehicle history (optional): every complaint ever filed for specific make + model + year via NHTSA's complaints API.
• New since last run mode with saved history per State key. An investigation that later closes is delivered again as a closed event.
• Privacy by design: VIN fragments, consumer city, dealer names/phones and other personal details are never output; phone numbers and e-mails typed into narratives are masked.
• Fast: a default run takes about 5 seconds at 512 MB.

Quick start

1. Click Start with the default input. You get about 40 records: NHTSA investigations opened or closed in the last 60 days plus the newest complaints received in the last 30 days, all makes.
2. Narrow it down: set makes, models, modelYears, a componentKeyword or tick crashOrInjuryOnly. Raise maxItems for bigger pulls (up to 20,000).
3. Daily defect alerts (schedule tip): create an Apify Schedule (e.g. every day at 9:00) with:

{ "mode": "sinceLastRun", "stateKey": "daily-tesla-ford", "makes": ["TESLA", "FORD"], "receivedWithinDays": 7, "investigationsWithinDays": 30, "maxItems": 2000 }

Each run delivers only complaints and investigation events not delivered before for that stateKey. Connect a webhook, Slack, Google Sheets or e-mail integration. Use one stateKey per schedule.

More examples:

• Air bag investigations of the last year: { "dataTypes": ["investigations"], "componentKeyword": "AIR BAG", "investigationsWithinDays": 365 }
• Fire/crash/injury complaints about fuel systems, last 90 days: { "dataTypes": ["complaints"], "componentKeyword": "FUEL", "crashOrInjuryOnly": true, "receivedWithinDays": 90, "maxItems": 500 }
• Every complaint ever filed for a 2024 Tesla Model Y: { "dataTypes": ["complaints"], "vehicleHistory": true, "makes": ["TESLA"], "models": ["MODEL Y"], "modelYears": [2024], "maxItems": 5000 }

Input

Field	Default	Meaning
dataTypes	both	complaints, investigations
makes	all	e.g. TESLA, FORD, HONDA
models	all	substring match: MODEL Y matches MODEL Y RWD
modelYears	all	e.g. 2023, 2024
componentKeyword	none	NHTSA component (or investigation subject) contains this text
crashOrInjuryOnly	false	only complaints with crash, fire, injury or death
receivedWithinDays	30	complaints received in the last N days (1-730)
investigationsWithinDays	60	investigations opened or closed in the last N days
maxItems	40	cap per run (also caps spend), newest first; investigations get up to half when both types are selected
vehicleHistory	false	all-time complaints for given makes + model years (+ models) via the NHTSA API
mode	window	window = everything in the window; sinceLastRun = only new (use for schedules)
stateKey	default	saved history name for sinceLastRun

Output (real records, Oct 2026; long text shortened)

Investigation that ended in a recall:

{
  "recordType": "investigation",
  "actionNumber": "PE24024",
  "investigationType": "PE",
  "investigationTypeName": "Preliminary Evaluation",
  "status": "closed",
  "dateOpened": "2024-09-06",
  "dateClosed": "2026-09-24",
  "latestEvent": "closed",
  "latestEventDate": "2026-09-24",
  "manufacturer": "Chrysler (FCA US, LLC)",
  "subject": "Underhood fires",
  "component": "STEERING:HYDRAULIC POWER ASSIST:PUMP",
  "recallCampaignNumber": "26V363",
  "makes": ["JEEP"],
  "models": ["GLADIATOR", "WRANGLER"],
  "modelYears": [2020, 2021, 2022, 2023, 2024],
  "summary": "On September 6, 2024, the Office of Defects Investigation (ODI) opened Preliminary Evaluation PE24024 to investigate allegations of engine compartment fires in model year 2021-2023 Jeep Wrangler and Gladiator vehicles...",
  "url": "https://www.nhtsa.gov/recalls?nhtsaId=PE24024",
  "recallUrl": "https://www.nhtsa.gov/recalls?nhtsaId=26V363000",
  "vehicleCount": 8
}

Complaint:

{
  "recordType": "complaint",
  "odiNumber": "11774329",
  "dateReceived": "2026-10-28",
  "incidentDate": "2026-08-26",
  "manufacturer": "Honda (American Honda Motor Co.)",
  "make": "HONDA",
  "model": "PILOT",
  "modelYear": 2020,
  "productType": "vehicle",
  "components": ["FORWARD COLLISION AVOIDANCE: ADAPTIVE CRUISE CONTROL", "FORWARD COLLISION AVOIDANCE: AUTOMATIC EMERGENCY BRAKING", "FORWARD COLLISION AVOIDANCE: WARNINGS"],
  "crash": false, "fire": false, "injuries": 0, "deaths": 0,
  "medicalAttention": false, "policeReport": false, "vehicleTowed": false,
  "incidentState": "NM",
  "complaintSource": "NHTSA website",
  "summary": "The front facing camera failed. Cruise control doesn't work. Emergency braking doesn't work...",
  "url": "https://www.nhtsa.gov/recalls?nhtsaId=11774329",
  "apiUrl": "https://api.nhtsa.gov/complaints/odinumber?odinumber=11774329",
  "similarComplaintsInWindow": 3
}

Full examples: sample-output.json. Every run also writes a summary (windows, filters, per-type counts, failures) to the OUTPUT record. The dataset has two table views: Complaints and Investigations.

Investigation types

PE = Preliminary Evaluation, EA = Engineering Analysis (an escalation), RQ = Recall Query (adequacy of a past recall), DP = Defect Petition (public request to open an investigation), AQ = Audit Query, EQ = Equipment Query, RP = Recall Petition.

Data sources, update timing and fair use

• NHTSA ODI complaints flat files (static.nhtsa.gov/odi/ffdd/cmpl/, e.g. COMPLAINTS_RECEIVED_2025-2026.zip), refreshed daily.
• NHTSA ODI investigations flat file (static.nhtsa.gov/odi/ffdd/inv/FLAT_INV.zip), refreshed daily; all investigations since 1972.
• NHTSA complaints API (api.nhtsa.gov/complaints/complaintsByVehicle) for the optional full vehicle history.
• U.S. Government works, free, no API key. Polite access with a declared User-Agent and retries.
• Complaints are unverified consumer reports; a complaint is not proof of a defect. Not affiliated with or endorsed by NHTSA or U.S. DOT. Confirm critical decisions against NHTSA's official records (url).
• Vehicle recalls are a different dataset: for FDA, NHTSA and CPSC recalls use our US Product Recalls Feed.

Pricing

Pay per event (Apify takes care of billing; you only pay for what is delivered):

Event	Price	Charged
apify-actor-start	$0.005	once per run
complaint-record	$0.003	per owner complaint delivered
investigation-record	$0.003	per NHTSA defect investigation delivered

A default test run (40 records: complaints + investigations) costs about $0.125. A daily sinceLastRun schedule for a few makes is typically a few cents a day. maxItems and Apify's max cost per run cap your spend; the Actor stops delivering when your budget is reached, so you are never charged for more than you set. No separate platform-usage charges: compute is included in the event prices.

FAQ

How fresh is it? NHTSA updates the flat files every day; complaints usually appear 0-2 days after NHTSA receives them. Can I get older complaints? Yes: raise receivedWithinDays up to 730, or use vehicleHistory for all-time complaints about a specific vehicle. Why do some complaints have several components? NHTSA codes one complaint against multiple components; they are merged into one record.