VIN Decoder & Recalls API - NHTSA Vehicle Lookup

Decode up to 1,000 VINs in one run and get the US safety recalls filed against each vehicle's make, model and year on the same row: make, model, year, trim, engine, body, assembly plant and safety equipment, the way the US federal record has them, plus every open recall filed against that make, model and year, with the fault, what it can lead to and the fix the manufacturer owes.

It is not a history report, though. Nothing about crashes, titles, mileage or owners, and no way to see whether a recall was already repaired on that particular car.

Input	17-character VINs, one per line
Output	One row per decoded vehicle: specification, plant, safety equipment and its open recalls
Ceiling	1,000 VINs per run
Account needed	None, and no API key
Price	$0.90 per 1,000 vehicles on the Free plan, a little less on paid plans. The free plan's $5 a month covers about 5,500

🔍 What VIN Decoder & Recalls API does

Each VIN is decoded against NHTSA's own vehicle record, the specification the manufacturer filed for that car.

Recalls are matched on make, model and year, because that is the level NHTSA publishes them at. So a campaign that only covers some build dates, one plant or one engine still shows up on every car of that model year. The summary names the range.

Model names do not always line up. The decoder calls a truck "Silverado" while the recall list files it as "SILVERADO 1500", "2500" and "3500". When the plain name finds nothing, the run tries the names NHTSA itself lists for that make and year, preferring the one that carries the VIN's own series, so a 1500 never picks up a 3500's recalls. recallModelMatched tells you which name matched.

And sometimes NHTSA's recall lookup returns nothing for a model its own index lists as recalled. In a check of 45 common US vehicles that happened on 5. Those rows say recallLookup: "unavailable" instead of claiming zero, so an unknown never passes for a clean record.

📋 What data you get from each VIN

What you get	Field
Make, model, year, trim and series	make, model, modelYear, trim, series
Body, vehicle type, doors, drive and fuel	bodyClass, vehicleType, doors, driveType, fuelTypePrimary
Engine and transmission	engineCylinders, displacementL, engineHorsepower, electrificationLevel, transmissionStyle, transmissionSpeeds
Manufacturer, plant and weight class	manufacturer, plantCountry, plantCity, plantState, gvwr
Safety equipment	abs, esc, tpms, airBagLocFront, airBagLocSide, airBagLocCurtain, seatBeltsAll
Open recalls: campaign, date, component, summary, consequence, remedy, park-it warnings	recalls, recallCount, recallLookup, recallModelMatched
Anything the decoder flagged about the VIN	nhtsaErrorCodes, nhtsaErrorText

▶️ How to decode VINs and check recalls in bulk

1. Open VIN Decoder & Recalls API and click Try for free.
2. Paste your VINs into VINs, one per line.
3. Set Maximum VINs to process to at least the number of VINs you pasted.
4. Leave Attach open safety recalls ticked, unless the specification is all you want.
5. Click Start, then export the dataset as CSV, JSON or Excel, filtered on charged.

Start it with no VINs and you get one free sample row showing the shape.

💰 How much does it cost to decode a VIN?

$0.90 per 1,000 vehicles on the Free plan, less on paid plans. That is $0.0009 per decoded vehicle, with recalls or without. The free plan's $5 a month covers about 5,500 vehicles.

Not charged: the sample row, values that are not VINs, VINs NHTSA has no vehicle for, duplicates in your list, and a vehicle whose recall lookup failed, which arrives free. A list of 400 VINs that all decode comes to $0.36.

📥 What you give it

{
  "vins": ["1HGCM82633A004352", "3GCUKREC5JG123456", "5YJ3E1EA7JF006588"],
  "includeRecalls": true,
  "maxItems": 25
}

Field	Default	What it is
vins	none; the box starts with three examples	Up to 1,000 VINs. Case, dashes and spaces do not matter, and duplicates are dropped before anything is charged. Left empty, you get the free sample row and no lookup.
includeRecalls	true	Attach open recalls. Turn it off for the specification alone; the price per vehicle is the same.
maxItems	25	How many VINs from the list get decoded, 1 to 1,000. The form starts at 25 too, so raise it when you paste more.
proxyUrls	empty	Optional, and only if you want the requests to go out through servers of your own, as http://user:pass@host:port. Leave it empty for a normal run.

📤 What you get back

A real row from a run on 30 September 2026, with the recall list cut to the first of its seven campaigns and the remedy text shortened:

{
  "ok": true,
  "charged": true,
  "recordType": "vehicle",
  "vin": "3GCUKREC5JG123456",
  "make": "CHEVROLET",
  "model": "Silverado",
  "modelYear": 2018,
  "trim": "LT",
  "bodyClass": "Pickup",
  "vehicleType": "TRUCK",
  "driveType": "4WD/4-Wheel Drive/4x4",
  "fuelTypePrimary": "Gasoline",
  "recallLookup": "ok",
  "recallCount": 7,
  "recalls": [
    {
      "campaignNumber": "19V645000",
      "reportReceivedDate": "2019-09-06",
      "component": "SERVICE BRAKES, HYDRAULIC:POWER ASSIST:VACUUM",
      "summary": "General Motors LLC (GM) is recalling certain 2015-2017 Cadillac Escalade, 2014-2018 Chevrolet Silverado, GMC Sierra, 2015-2018 Chevrolet Suburban, Tahoe, GMC Yukon vehicles.  The amount of vacuum created by the vacuum pump may decrease over time.",
      "consequence": "As the vacuum level drops, the brake assist decreases, increasing braking effort, extending the distance required to stop the vehicle, thereby increasing the risk of a crash.",
      "remedy": "GM will notify owners, and dealers will reprogram the Electronic Brake Control Module, free of charge.  The recall began October 8, 2019. ...",
      "notes": "Owners may also contact the National Highway Traffic Safety Administration Vehicle Safety Hotline at 1-888-327-4236 (TTY 1-800-424-9153), or go to www.safercar.gov.",
      "manufacturer": "General Motors LLC",
      "parkIt": false,
      "parkOutSide": false,
      "overTheAirUpdate": false
    }
  ],
  "series": "1500",
  "doors": 4,
  "engineCylinders": 8,
  "displacementL": 5.3,
  "engineHorsepower": null,
  "electrificationLevel": null,
  "transmissionStyle": null,
  "transmissionSpeeds": null,
  "manufacturer": "GENERAL MOTORS LLC",
  "plantCountry": "MEXICO",
  "plantCity": "SILAO",
  "plantState": null,
  "gvwr": "Class 2F: 7,001 - 8,000 lb (3,175 - 3,629 kg)",
  "abs": null,
  "esc": null,
  "tpms": "Direct",
  "airBagLocFront": "1st Row (Driver and Passenger)",
  "airBagLocSide": "1st Row (Driver and Passenger)",
  "airBagLocCurtain": "All Rows",
  "seatBeltsAll": "Manual",
  "recallModelMatched": "SILVERADO 1500",
  "nhtsaErrorCodes": "1",
  "nhtsaErrorText": "1 - Check Digit (9th position) does not calculate properly",
  "vehicleDescriptor": "3GCUKREC*JG",
  "scrapedAt": "2026-09-30T02:37:59.642Z"
}

Field	How to read it
recallLookup	Read this before recallCount. ok: the recall list answered and the campaigns are attached. none: it answered and has no campaign for this vehicle. unavailable: the count is unknown, not zero. skipped: you turned recalls off.
recallCount, recalls	How many campaigns apply, and each one in full, with the "park it" and "park outside" warnings as parkIt and parkOutSide. The list stops at 60; recallCount still counts them all.
recallModelMatched	The model name the recall list matched. Usually the same as model.
modelYear	The model year, which is not always the year the car was built.
trim	Sometimes a slash-separated list, when one VIN pattern covers several trims.
nhtsaErrorCodes, nhtsaErrorText	Whatever the decoder flagged. Code 1, a check digit that does not add up, is common on VINs that decode perfectly well.

Any field the manufacturer never filed comes back as null. Older and low-volume vehicles have more of those than a recent mass-market car, and running them again does not change that.

🧾 Reading the output

Row	How to spot it	Charged
A vehicle	charged: true, recordType: "vehicle"	yes
The sample row	_sample: true, charged: false. Only when the run had no VINs at all	no
A diagnostic row	_diagnostic: true, charged: false, and an errorCode	no

The sample row carries recordType: "vehicle" as well, so filter on charged, not on recordType, to keep only the vehicles you paid for. Read charged as that marker, not as a receipt.

errorCode	What happened
BAD_INPUT	The value is not 17 characters, or it contains I, O or Q, which no VIN does.
NO_RESULTS	NHTSA holds no vehicle for that VIN, which is common for cars never sold in the US, or it read the VIN as incomplete.
RECALL_LOOKUP_FAILED	The vehicle decoded but its recalls could not be fetched. The whole vehicle is in the row anyway, free. Run that VIN again later for its recalls.
NETWORK	The decoder did not answer for that group of VINs.
TIME_BUDGET	The run hit its time limit before it reached that VIN, or before it looked up its recalls. Nothing was charged for it. The run ends as failed only when the time limit stopped it before any vehicle came back.

On a long list of bad VINs, the first 50 failures get a row each and the rest are collected in one NO_RESULTS row whose skipped list names every VIN and its reason. If 150 VINs in a row produce no vehicle at all, the run stops and says so, since the list is clearly not VINs.

💡 What people use it for

• A lot full of trade-ins, where an open airbag or fuel-pump campaign should turn up before the car is listed, not after a buyer finds it.
• A fleet checked every week, where only the new campaigns matter: compare recallCount and the campaign numbers against the last run.
• A listings feed where sellers type the model any way they like, and the specification should come from the VIN instead.
• A book of VINs for claims or rating work that needs make, model, year, body and safety equipment without anyone retyping them.

🚧 What it does not do

• No history. No accidents, titles, salvage brands, odometer readings, thefts or service records. None of that is in the federal record.
• No owners. Nothing about who owns or registered the car, and no field that would ask for it.
• No per-car recall status. A campaign that applies is not the same as a repair still owed. Whether the work was done is known only to the manufacturer and its dealers.
• US vehicles only. A car never certified for US sale decodes partly or not at all, and has no recall record here.
• Nothing before 1981. Older VINs are not 17 characters and are refused.
• Only federally filed recalls. Recalls issued outside the US, and service campaigns never filed with NHTSA, are not included.
• At most 60 campaigns listed per row, and 1,000 VINs per run. Split longer lists across runs.
• Not instant on mixed lists. Recalls are fetched once per make, model and year, so a list of the same few models is quick and hundreds of different models take minutes.
• Not an official NHTSA tool: Dami's Studio is independent and is not affiliated with or endorsed by NHTSA.

🧭 Which vehicle data scraper do you need?

If you want	Use
The factory specification and open US recalls for a VIN	This one
Used-car listings in eight European countries	AutoScout24 Scraper
Car listings in Brazil, priced against the FIPE table	Webmotors Scraper
Pump prices at UK filling stations	UK Fuel Prices Scraper

❓ Questions people ask

Is this a vehicle history report?

No, and it cannot be made into one. If the car's past is what you need, buy a history report.

Does a recall count of 0 mean the car is clean?

Only when recallLookup says none, and that covers recalls filed with NHTSA for that make, model and year, nothing more. When it says unavailable, the honest answer is that nobody knows yet.

Why does a row say the check digit does not calculate?

That is the decoder's code 1, and it is common on VINs that decode perfectly well. It often means one character was copied wrong, so read the VIN again, but the row is still a real decode.

What if a VIN is wrong or made up?

You get one free row saying why, and the run carries on with the others.

Do I need an NHTSA API key or an account?

No. Paste the VINs and start. Nothing is kept between runs, so the same list every week on a schedule is safe.

Can I call it from code or connect it to an AI assistant?

Yes. The API tab has ready-made code for Python, JavaScript and the command line. For Claude, ChatGPT or another MCP client, connect https://mcp.apify.com/?tools=fetch-actor-details,dami_studio/vin-decoder-recalls-scraper. Either way the run happens on your Apify account at the same price.

Is it legal to look up VINs?

Both the specification and the recalls come from records the US government publishes. Apify's write-up on the legality of web scraping is a good starting point, and we are not lawyers.

🆘 If something breaks

Open the Issues tab on the actor page. Include the run ID and the VINs that went wrong, with the errorCode from their rows.