Vehicle History

How to Check if a Car Is Stolen Before You Buy It

A person holding a phone beside a car, checking something before buying

The short version

  • There is no single database of stolen cars that the public can search. There are several, they are run by different organisations for different reasons, and each one has a hole in it the others do not.
  • The free check most people find first — NICB’s VINCheck — only sees what its member insurance companies have reported. A car stolen from a private owner who never filed an insurance claim can be absent from it entirely.
  • Every one of these checks answers the same narrow question: has this seventeen-character string been reported? None of them can tell you that the string belongs to the car parked in front of you.
  • That gap is the whole problem. A VIN swap defeats every database check ever devised, and it is defeated in turn by two minutes of comparing plates on the car itself.
  • If a car turns out to be stolen after you buy it, you generally lose both the car and the money. A thief cannot pass good title, and paying in good faith does not fix that.

The car is a good one. The price is a little under what you expected, which the seller explains easily enough. They want cash, they want to meet somewhere other than the address on their licence, and they are relaxed and friendly about all of it.

Almost everybody who searches this phrase is standing in some version of that moment. They do not want a lecture on auto theft. They want to know, in the next few minutes, whether the car is clean — and they have found three websites that promise to tell them, all of which return a green tick.

Here is what those green ticks actually mean, in what order to run the checks, and the one verification that matters more than all of the databases combined — because it is the only one a competent thief cannot arrange in advance.

Why there is no single “is it stolen” database

The intuition behind the search is that somewhere there is a master list, and the job is to find a website with access to it. That list does not exist in a form you can search, and understanding why saves a great deal of misplaced confidence in the answers you do get.

A stolen vehicle generates records in several places at once, and those places have never been merged. The police agency that took the report enters it into a law-enforcement system that is not open to the public. The insurer that paid the claim files it with the industry’s own crime bureau. The state that eventually retitles the vehicle records a brand on the title. A commercial history company buys what it can from whoever will sell it.

Each of those organisations is answering its own question. Law enforcement wants to recover the vehicle and prosecute somebody. The insurance industry wants to detect fraudulent claims. The state wants the title record to be accurate. None of them was designed to give a private buyer a yes-or-no answer on a Sunday afternoon, and the tools that appear to do so are all reading one slice of one system.

The practical consequence. A clear result from any single check means “this VIN did not appear in the one dataset I can see”. It does not mean the car is clean. Running three checks that draw on three different datasets is not paranoia — it is the only way to cover the gaps, and it still takes under ten minutes.

A police patrol car parked on a city street
The report that starts the record is filed with a police agency, in a system the public cannot search. Everything a buyer can reach is downstream of it.

The three checks worth running, in order

Run these in sequence. The order matters, because each one is cheap and fast in proportion to how much it rules out, and because the last of them is the only one that examines the car rather than the string.

One: the federal decoder, to confirm the VIN is real

Before asking whether a VIN has been reported, establish that it is a VIN at all. NHTSA’s public decoder takes the seventeen characters and returns what the manufacturer encoded into them — make, model, year, body style, engine, plant. It is free, it has no search limit, and it is the fastest way to catch a string that was invented rather than stamped.

Two things to watch for. First, the specifications it returns must match the car you are looking at. A VIN that decodes to a four-cylinder saloon while you are standing beside a six-cylinder coupé has already answered your question. Second, and much less well known, the decoder returns an error line when the check digit in position nine does not calculate — a mathematical test of the other sixteen characters that catches invented and mistyped VINs. Most consumer tools strip that error out before showing you anything. Our guide to the VIN year chart and the check digit works through running that arithmetic yourself.

Two: NICB VINCheck, for the insurance record

The National Insurance Crime Bureau is a non-profit funded by the insurance industry, and it runs a free public lookup called VINCheck. Enter a VIN and it reports whether that vehicle has been recorded as stolen and not recovered, or as a salvage or total-loss vehicle, by participating member insurance companies.

That qualifier carries the entire weight of the tool’s limitations, and the next section is about it. The service is free and allows a limited number of searches from one address in a day — enough for a buyer looking at one or two cars, deliberately not enough to run a list.

Three: the title record, through NMVTIS

The National Motor Vehicle Title Information System is run by the US Department of Justice, and it is the one federal system that pulls state title records together. Its consumer-facing reports come from approved providers rather than from the government directly. What it shows you is the title history — including brands such as junk, salvage and flood, and in many states a theft record that made it onto the title.

This is a different slice of the problem from VINCheck. NICB sees insurance claims. NMVTIS sees state title documents. A car can appear in one and not the other, which is exactly why both are on this list. We cover what NMVTIS does and does not contain in the guide to the federal record behind the commercial reports.

What NICB VINCheck covers, and what it misses

VINCheck is genuinely useful and genuinely free, and it is presented across the internet as more comprehensive than it is. The service reports what participating member insurers have filed. Work through what that excludes.

A car is stolen from a private owner. The owner carried liability-only cover, as a large share of owners of older cars do, so there is no theft claim to file — the insurer owes nothing for the vehicle. A police report exists. An insurance record does not. VINCheck has nothing to report because nothing was ever reported to the industry that feeds it.

Or: the theft is recent. The claim has been filed but not yet settled and propagated into the shared dataset. The car is on the market during exactly that window, because a thief’s whole business model is speed.

Or: the insurer is not a participating member. Coverage is broad but it is not universal, and the tool tells you it is reporting member data rather than all data.

What a clear VINCheck result actually licenses you to believe. That no participating member insurer has filed a theft, salvage or flood record against this VIN. That is a real and useful fact. It is not the same fact as “this car was not stolen”, and the distance between the two is where buyers get hurt.

The VIN on the car versus the VIN in the database

Everything above shares one silent assumption: that the seventeen characters you typed describe the vehicle you are standing next to. Strip that assumption away and every database check on the internet becomes worthless simultaneously.

The technique is old and it is simple in concept. Find a legitimate vehicle of the same make, model, year and colour — one sitting in a salvage yard, or one you can photograph in a car park. Take its identity. Apply that identity to the stolen car by replacing or altering the visible VIN plate and the door label. The stolen vehicle now presents the identity of a car with a clean history, and it will pass every lookup, because the lookups are being run against the clean car’s identifier.

This is why “I ran the VIN and it came back clean” is a weaker statement than it sounds. It is a true statement about a number. Whether it is a true statement about a car depends entirely on whether the number came from the car — and that is a physical question, answered with your eyes, not a database question.

Checking the car’s own identity plates agree

A modern vehicle carries its VIN in several places by design, and the reason is precisely this. Duplicating the identifier across the vehicle means that altering one instance leaves the others disagreeing with it. The check is to read them all and confirm they match.

Start at the base of the windscreen on the driver’s side, where a small plate is visible from outside the car. Then open the driver’s door and read the label on the door jamb or the door edge, which carries the VIN alongside tyre pressures and a manufacture date. Depending on the vehicle, the number is also stamped into structure under the bonnet, and printed on the registration and insurance paperwork the seller hands you.

Our full guide to where the VIN is located on a car covers each position in detail. For this purpose what matters is not finding one — it is confirming that every one you find reads identically, and that none of them looks like it has been interfered with.

What tampering actually looks like

You are not looking for craftsmanship. You are looking for the absence of factory consistency.

  • Rivets that do not match. The windscreen plate is factory-fitted with rivets of a particular type and finish. Screws, adhesive, or rivets that look newer than the car are all wrong.
  • A door label that is too clean, or peeling. The jamb label ages with the car. One that looks freshly printed on a vehicle with worn paint and a shiny label edge deserves a second look. Bubbling, lifted corners and misalignment with the panel are all signs it has been off and back on.
  • Character spacing that drifts. Factory stamping and printing is machine-even. Hand-altered characters wander — uneven gaps, inconsistent depth, a character sitting slightly high or low against its neighbours.
  • Scratches, filing or fresh paint around a plate. Any disturbance to the surrounding metal or trim on a car with no accident history in that area is a question that needs an answer.
  • A plate that is simply missing. On a car of an age that should have one, an absent VIN plate is not an oddity to work around. It is the end of the viewing.
A person sitting in a car checking something on a smartphone
The database check and the physical check answer different questions. Running the first without the second tells you about a number, not about a car.

Reading the paperwork and the seller

Theft cases that reach buyers tend to share a shape, and it is usually visible in the documents and the circumstances well before it is visible in the metal.

The name on the title is not the name of the person selling. This is the single most common pattern. There is always an explanation — selling for a relative, for a friend, for an estate, a dealer who has not yet retitled. Some of those explanations are true. The response to all of them is the same: the sale happens with the titleholder present, or it does not happen.

The title is fresh, and from another state. A recently issued title from a state the seller has no connection to is how a brand gets washed and how a history gets shortened. It is not proof of anything on its own, and it is a reason to pull the title record before going further.

Only one key exists. Modern keys are expensive to replace and owners generally have both. A single key on a car that should have two is consistent with a car whose owner never handed the keys over.

The seller resists the ordinary friction of a legitimate sale. No inspection at a garage of your choosing. No meeting at the address on the registration. Cash only, today, and a reason why tomorrow will not work. Each of these has an innocent version. Together they are a pattern, and the pattern is the point.

The test that costs nothing. Ask to meet at the seller’s home address, the one printed on the title, and ask for a pre-purchase inspection at a garage you pick. A legitimate private seller finds both requests mildly inconvenient. Somebody selling a car they do not own finds a reason why neither is possible.

What to do if you think a car is stolen

Do not confront the seller. There is no version of that conversation that ends with you better off, and you have no way of knowing who you are dealing with.

Leave. Say the car is not right for you, or that you want to think about it, and go. You owe a stranger nothing beyond that, and no deposit is worth staying to argue about.

Then write down what you have: the VIN, the registration plate, the address where you met, the phone number, a description of the seller and the vehicle, and any photographs you took while viewing. Photographs of the VIN plate and door label are particularly worth having.

Report it to the police non-emergency line for the area where you saw the car. They can query systems you cannot. Give them what you wrote down and let them establish whether the vehicle is in fact reported — that determination is theirs to make, and it is genuinely useful to them even when it turns out to be nothing.

What happens if you have already bought one

This is the part that people find hardest to believe, and it is the reason the checks above are worth doing carefully rather than quickly.

A thief has no ownership to transfer. Because they never held good title, they cannot pass good title to you, and you cannot pass it to anybody else. When a stolen vehicle is identified it is generally recovered and returned to its rightful owner or that owner’s insurer. The buyer is left holding a bill of sale from a person who has usually become impossible to find.

Good faith does not change the outcome. Paying market price, keeping receipts, and having no idea the car was stolen are all true of most people this happens to, and none of it creates ownership that the seller never had to give. Your recourse runs against the seller, which is worth exactly as much as your ability to locate them.

If you find yourself here: report it immediately rather than waiting, keep every document from the purchase, and get advice about your position. Reporting it promptly is also what separates a victim from somebody who looks like a participant.

Where theft records and title brands overlap

A stolen car that is recovered does not simply return to circulation unmarked. If it was damaged, stripped or burned before recovery, the insurer that already paid the claim owns it and disposes of it — and it typically re-enters the market carrying a brand.

This is why VINCheck reports theft, salvage and flood together rather than treating them as separate services: they are stages of one process. The record you find may not say “stolen” at all. It may say salvage, because the theft ended in a total loss and the brand is what survived into the title record.

The practical reading is that a brand on a recovered theft vehicle tells you nothing about the quality of any subsequent repair. What that brand does mean, and what it costs a buyer, is covered in our guide to what a salvage title actually records.

Frequently asked questions

Can I check if a car is stolen for free?

Yes, partly. NICB’s VINCheck is free and reports theft and total-loss records filed by participating member insurers, with a small daily search limit per address. NHTSA’s decoder is free and unlimited, and confirms the VIN is structurally valid and matches the vehicle. Neither is a complete theft check, and the free portion of the job is genuinely useful — it is just not the whole job.

Does a clear VINCheck result mean the car is definitely not stolen?

No. It means no participating member insurer has filed a record against that VIN. A theft never claimed on insurance, a theft too recent to have propagated, or a theft insured by a non-member company can all produce a clear result. It is evidence, not proof.

Will a paid vehicle history report tell me if a car is stolen?

A paid report draws on more sources than any single free check, including the federal title record, and it is a reasonable purchase before committing thousands of pounds or dollars. It has the same structural limitation as every other database: it reports on the VIN you give it. It cannot verify that the VIN belongs to the car in front of you.

What if the VIN plate is missing or looks altered?

Walk away. A missing, damaged or apparently altered VIN plate on a vehicle that should carry one is not an administrative problem to sort out after purchase — it is the specific condition every one of these checks depends on being absent. There is no price that makes it worth taking on.

Can the police tell me whether a car is stolen before I buy it?

They can query systems that are not open to you, and a non-emergency line is the right place to take a genuine suspicion with a VIN and a location. Treat it as a route for reporting something that looks wrong rather than as a routine pre-purchase service, and do the free checks yourself first.

Is buying a stolen car a crime if I did not know?

Buying without knowledge is not the same as knowingly receiving stolen property, and the distinction is exactly why prompt reporting matters. Someone who reports as soon as they find out, with a bill of sale and a record of the checks they ran, is in a very different position from someone who kept driving it. This is general information rather than legal advice; if it has happened to you, get advice about your own circumstances.

How long does it take to run these checks properly?

The three database checks take under ten minutes between them. The physical comparison of the VIN plates takes about two, and it is the one that catches the case the databases cannot see. There is no version of this that is too slow to be worth doing before handing over money.

Sources and further reading

Recall, complaint and safety-rating figures on this page were retrieved from the federal databases above on August 19, 2026. Federal data changes — re-check any VIN before you rely on it.

Baron Auto Editorial Team We research used cars against federal data — NHTSA recall campaigns, owner complaints and EPA fuel-economy records — and publish what we find. We do not sell cars, loans, or insurance, and no manufacturer or dealer pays for coverage here.

Last updated August 21, 2026. Found something out of date or wrong? Tell us and we will correct it.