What a VIN encodes, how the check digit is computed, what the NCIC codes are for, and what this site's data does and does not cover.
A 17-character Vehicle Identification Number defined by ISO 3779. It encodes the manufacturer, vehicle attributes, model year, assembly plant, and a serial number.
No. Every VIN generated here uses a randomised serial, and none of them is taken from a real-vehicle record — we never publish VINs from real-vehicle datasets. They are structurally valid and decode correctly, but what they describe is a pattern, not a vehicle. A random serial can still land on a real VIN by chance.
The first 11 characters follow genuine manufacturer patterns from NHTSA's VPIC database, so decoders read them correctly. Only the serial is invented.
Extremely unlikely, but not impossible. Real serials are typically sequential; ours are random across a 33-character alphabet. Never treat a generated VIN as evidence about a real vehicle.
Testing forms and validation, seeding test fixtures, demos, screenshots, training material, and load-testing decoders.
Anything that represents a generated VIN as a real vehicle's identifier: titling, registration, insurance claims, listings, or VIN cloning. That is fraud in every US jurisdiction.
Position 9. It is computed by transliterating the other 16 characters, applying positional weights, and taking the total modulo 11, where a remainder of 10 becomes the letter X. It catches transcription errors. There is a full worked example on the check-digit page.
The model year. The codes repeat on a 30-year cycle, so A is both 1980 and 2010; other positions disambiguate which is meant.
Usually because its WMI is absent from VPIC — common for low-volume, imported, or pre-1981 vehicles — or because VPIC does not cover that pattern. We return whatever is derivable from the VIN plus the specific reason.
The data comes from NHTSA's VPIC database, which covers vehicles sold or registered in the United States. A brand never sold here — Škoda, Scania, most Chinese marques — is not in it, and neither is any pre-1981 vehicle, which predates the 17-character VIN.
Standardised FBI/NCIC codes for make, colour, and style — TOYT, BLK, 4D — that appear on titles, registrations, and law-enforcement records. Title and DMV work needs them, and no other VIN decoder outputs them.
Yes. Use the bulk generator, or the API's make, model, year, bodyStyle and fuelType parameters.
Up to 100 per request, both in the browser and through the API.
Yes. Pass a seed, and the same seed returns the same batch every time.
US-market vehicles, model years 1980 to 2026, focused on passenger cars, trucks and MPVs. Trailers and low-speed vehicles are out of scope. Vehicles never sold in the United States are not in VPIC and so are not here either.
Yes. No signup, no key, and no quota beyond a per-IP rate limit that exists to keep the service available.
No. The service is stateless with no database, and decode requests are not logged with their VINs.
NHTSA's VPIC database, by way of @cardog/corgi for decoding, and NIEM 6.0 for the NCIC codes by way of ncic-vehicle-lookup.