Every figure names its source. This page says where it stops.
What is in it, where it came from, and what it is not
Every figure on this site is traceable to a federal record or a manufacturer document. This page says which, how often each one moves, and where the coverage stops.
Where each field comes from
| Layer | Source | Refresh | Coverage |
|---|---|---|---|
| Fuel economy, engines, configurations | U.S. EPA certification records | Weekly | Every model year |
| Recalls, defect text, build windows, completion rates | NHTSA, including 49 CFR 573 filings | Nightly | Every model year |
| Owner complaints, failure mileage | NHTSA complaint database | Nightly | Every model year |
| Crash ratings | NHTSA New Car Assessment Program | As published | Rated model years only |
| Service bulletins and campaigns | NHTSA manufacturer communications | Nightly | Most model years |
| Diagnostic trouble codes | WhichTrim, derived from the federal record | With the build | Where the record supports one |
| Engine, plant, filed equipment | NHTSA vehicle product information | Monthly | Most model years |
| Trim specs, equipment, MSRP | The manufacturer's own press kit or newsroom spec sheet | On publication | 1,108 model years so far |
| Measured dimensions, curb weight | Transport Canada, Canadian Vehicle Specifications (Open Government Licence – Canada) | Weekly | 2,414 model years without a press kit; a figure outside physical bounds is left out, not corrected |
| Fuel, diesel and electricity prices | U.S. Energy Information Administration | Weekly / monthly | All 50 states and DC |
| Public EV charging sites | Federal alternative-fuel station register | With the build | All 50 states and DC |
Every page also publishes the JSON it was built from, so none of this has to be taken on trust. The methodology page has the arithmetic.
What this is not
Stated here rather than left for you to discover. A dataset is only as useful as its edges are clear.
Coverage starts at 2016.
U.S.-market cars, trucks and SUVs only. There is no page for a 2009 Accord and we would rather say so than build a thin one.
Complaints are unverified, and they scale with sales.
They are what owners told NHTSA, not adjudicated defects. A best-seller will always show more of them than a rare car with the same problem, which is why they are never folded into the Safety Signal grade.
Equipment recalls are not here — only vehicle campaigns.
Every campaign on this site is one an automaker filed against a vehicle. NHTSA also records recalls filed by parts makers against a component — a tyre, a child seat, an aftermarket accessory — and those are not listed or counted anywhere here. NHTSA shows an owner both, so a VIN check at nhtsa.gov/recalls can return something this site does not have. We would rather name the edge than quietly sit inside it.
A recall covers a build range, not a model year.
Nothing on this site — and nothing in the public record — can tell you whether one specific vehicle is affected: the list of VINs inside a campaign exists only at the manufacturer. So every vehicle page carries both doors that do settle it — nhtsa.gov/recalls and that make's own lookup, which 49 CFR 573.15 requires it to run and to keep current to within seven days.
Trim-level specs exist only where a press kit does.
Manufacturer trim data exists for 1,108 model years. The rest have EPA and NHTSA data, plus the equipment the manufacturer filed with vPIC and, where Transport Canada measured the model, its dimensions and curb weight. A page tells you which it is showing you.
We hold no ownership, title, odometer or accident history.
Not a little of it — none. That is a different industry with different licences, and a dataset that guesses at it is worse than one that does not have it.
Nine states have their own gasoline price. The rest are regional.
That is what EIA publishes, so every regional figure on the site is marked as one rather than presented as a state price.
What we guarantee
Five of them, narrow enough to be worth something. Each one is a constraint we build under, not a policy we publish — which is why they can be checked from the outside.
Every value names its source.
Not a bibliography at the bottom — the record each number came from, next to the number, on the page. It is the reason this page can be short.
A public safety record is never behind a payment.
Every recall, complaint and crash rating on this site is free to read and always will be. What a paid plan buys is the watching, the joins and the delivery — never access to a record the public already owns.
We never store or publish a full VIN.
Where NHTSA's complaint data carries one, it is truncated to the nine characters that describe the configuration before anything is written down. The check digit, plant code and serial are dropped, so nothing here identifies a vehicle or the person who complained.
No personal data is sold. Not now, not ever.
We hold your name, email, sign-in id and usage counts — and, if you use the Garage, the cars you add, what you note about them and any files you keep in its Glovebox. None of it is for sale at any price, and none of it is shared with advertisers or data brokers.
When a source is wrong, we inherit it — and say so.
We do not silently correct federal records. Every page links the JSON it was built from so you can see exactly what we read, and the build re-reads the sources daily.
Every one of these is visible in the product: the sources sit next to the numbers, the safety pages carry no paywall, and the JSON behind each page shows exactly what was read.
The data behind this
Everything above, as files. No key, no form, no attribution beyond the line below — if it is useful, use it.
How to cite
WhichTrim, “WhichTrim: US vehicle recalls, complaints, crash ratings and trim specs by model year”, 2026-11-04. https://whichtrim.com/coverage/
The underlying records are U.S. federal public-domain data from NHTSA. The joins, the filtering and the arithmetic are ours, and we would rather be credited than not — but the figures are checkable from the source either way, which is the point.
What you may do with it
Read it, quote it, cite it. Build on it on a paid plan. The federal records underneath are public domain and yours already; the joins, grades, transcriptions and identifiers are ours and licensed to you for the life of your plan.
If you are publishing a figure from here, the attribution we ask for is one line: Data: WhichTrim, from EPA and NHTSA public records.
The underlying federal records are public domain and we claim nothing in them. The compiled dataset — its selection, structure and arrangement, together with the derived fields, grades, transcriptions and identifiers created by WhichTrim — is our proprietary work product. A plan licenses its use; it does not transfer ownership, and it does not permit extracting the dataset to build a substitute for it.
Questions buyers actually ask
Source: EPA, NHTSA and our own buildCan I use this commercially?
Yes, on a paid plan, inside your own product or analysis — see the API terms for the exact wording. The free plan is for non-commercial use and asks for attribution. What no plan allows is reselling the dataset itself or using it to build a substitute feed.
Where does the federal data end and yours begin?
The records are public domain and you do not need us to get them. Everything between those records and what you see here — the joined dataset, the grading, the transcriptions, the derived fields and the identifiers — is WhichTrim's own work product, and it is what a plan licenses.
How current is it?
The site rebuilds daily against the current federal files, and every page carries the timestamp it was generated at. Manufacturer figures can lag a mid-year change, because they are only as current as the document the manufacturer published.
What happens if I find something wrong?
Check the JSON linked at the foot of the page first — it shows exactly what we read. If we misread it, that is ours to fix. If the source is wrong, we will not quietly overwrite a federal record, but we will say so on the page.
Is there an uptime commitment?
Business plans carry a 99.9% monthly uptime target with service credits. The static JSON twins of every page are served from a CDN and have no practical rate limit.