Carfax Odometer Check — Where Every Mileage Reading Actually Comes From.

Yes. A Carfax vehicle history report includes reported odometer readings presented with the date and the source of each entry, and that presentation is the point — one mileage figure on its own cannot be verified against anything, while a dated sequence of them can be tested for direction and rate. What the report contains is not the car's mileage in any absolute sense; it is the set of readings that happened to be recorded by someone who reports into a system. Readings enter from several distinct places: the certification a seller signs when a state issues a title, state safety and emissions inspections, service and repair visits, wholesale and salvage auctions, and insurance events such as a total-loss valuation. Carfax's particular strength is the service-visit layer, which it has been building from franchise dealers and service chains since 1984. Its weakness is the same as every provider's — a car serviced by an independent garage and sold privately for cash can travel a long way without generating a single reportable reading.

They overlap substantially at the foundation and diverge at the edges. The foundation is NMVTIS, the National Motor Vehicle Title Information System operated by the US Department of Justice, which records the odometer reading disclosed when a state issues a title along with later mileage where available. Every legitimate provider draws on it, which is why the title-issuance readings and the brand records tend to match across reports — they are the same federal data. The divergence comes from the proprietary layer each provider adds on top. Carfax contributes readings from its own service-record network, so on a car maintained at franchise dealers it will often show more intermediate data points. An NMVTIS-backed report concentrates on the records with legal weight behind them: the certified disclosure at each transfer and any not-actual-mileage brand a state has applied. For the specific question of whether mileage has been rolled back, the federal layer is the one that settles it, because a dip between two certified readings is decisive regardless of how many service entries sit around it.

Carfax does not sell an odometer check separately — the mileage history is a section within the full vehicle history report, so the report price is the price. A single Carfax report is about $44.99. AutoCheck sells a single report at $29.99, VINAudit at around $9.99, and a CarCheckerVIN full report is $14.99, with a 10-pack at $89.99 working out to $9.00 per report on credits valid for twelve months. The multi-report pricing matters more than it first appears, because odometer checking is rarely something you do once: buyers usually screen several candidates and abandon most of them. Paying a single-report price per car makes screening expensive enough that people skip it on the cars they are less sure about — which is exactly backwards. The practical approach is to establish free whether a mileage problem exists, then spend on the full record for the one or two cars that survive that filter.

Yes, and it is the right first move regardless of which provider you eventually buy from. The free check on this page takes the 17-character VIN and returns the reported odometer readings along with any not-actual-mileage title brand, which is enough to resolve the binary question — is there a mileage problem here or not. That is genuinely all you need to walk away from a bad car, and walking away is the outcome that saves the money. What the free tier does not give you is the surrounding detail: the full ownership chain, every reported accident, and the complete record behind each entry. Carfax has no free full report, so the comparison is not free-versus-paid but rather how much you spend to answer a question you may be able to answer for nothing. Buy the deeper report when the timeline is ambiguous or the car is one you actually intend to purchase, not as a first step.

It means nobody recorded a reading during that period, which is an absence of observation rather than evidence of wrongdoing — and confusing the two is the most common mistake people make with a mileage timeline. Gaps are normal and have mundane explanations. A car can stay with one owner for eight years without a title being issued, and no federal disclosure is triggered until it changes hands. It can be maintained by an independent workshop that reports into no network. It can live in a state that does not capture mileage at inspection, or be exempt from disclosure entirely: under 49 CFR 580.17 a vehicle from model year 2010 or earlier becomes exempt ten years after January 1 of its model year, while a 2011-or-later vehicle is exempt only after twenty years. What a gap should do is direct your attention rather than settle the matter. Look at the readings either side and ask whether the accumulation across the interval is plausible for the time elapsed. A gap bracketed by two consistent readings is uninteresting; a gap that coincides with a change of state and a suspiciously flat rate is worth a question.

The one captured when a state issued a title, because it is the only category with a signed federal certification behind it. Under 49 CFR 580.5 the transferor certifies in writing that the reading is the actual mileage, or that it exceeds the designed mechanical limit, or that it does not reflect actual mileage — and the last of those must carry a warning notice that a discrepancy exists. That signature converts the figure from a data-entry event into a legal statement, which is what makes it both more reliable and more useful if things go wrong later. Insurance readings rank well for a different reason: a total-loss valuation drives a payout, so someone had a financial motive to record the number accurately. Service and inspection readings are useful for filling gaps and establishing rate, but they are typed by hand at a counter and transposition errors do happen, which is why a single anomalous service reading is worth checking rather than acting on. The general principle is to weight readings by what was at stake for the person recording them.

Separate the two situations first, because they call for different responses. If a later reading sits below an earlier one, that is not an anomaly to investigate — it is the finding, since no legitimate mechanism reduces cumulative distance. Preserve everything before you act: photograph the cluster, the pedal rubbers, the seat bolster and the steering wheel, and keep the advertisement, the bill of sale and the signed odometer disclosure together, because servicing or cleaning the car destroys physical evidence. If instead the sequence rises but a stretch looks implausibly flat, you have a question rather than a conclusion, and the seller may well have an ordinary answer involving a car that sat unused or a replaced instrument cluster with an invoice to prove it. Where you have bought the car already, note that 49 U.S.C. § 32710 makes a violation committed with intent to defraud worth three times actual damages or $10,000, whichever is greater, requires the court to award costs and a reasonable attorney's fee on judgment, and requires the action to be brought within two years of the claim accruing. That is general information, not legal advice.

Where a state has formally branded the title, effectively yes, and this is one of the more useful features of the system for buyers. Brands report into NMVTIS, so they travel with the VIN across state lines rather than disappearing when the vehicle is retitled somewhere new — which is precisely the manoeuvre, title washing, that branding exists to defeat. A car carrying a not-actual-mileage brand keeps it, and anyone running the VIN can see it several owners later. The corollary is worth stating plainly, though, because it is where people over-read a clean result: the absence of a brand is not a certificate of honesty. It means no state has yet reached a formal finding, which may be because nothing is wrong or may be because nobody has looked. So read the two things together — the brand record tells you what an authority has already concluded, and the reading sequence tells you what the numbers themselves say. Either can flag a problem the other misses.