Numbers nobody else has joined up

Findings from the public federal record and the manufacturers' own press kits, with the method stated, the caveats stated, and the whole table attached as a file. If you disagree with one, you can check it with our own data.

Photo: Wesley Tingey / Unsplash

Why we publish the file

Anyone can put a big number in a headline. The number is only worth anything if you can check it.

So every study here comes with the whole table behind it, as a CSV and a JSON file: which records went in, which were left out and why, and a link from every row to the federal filing or press kit it came from. Add it up yourself and you should land on our figure. Where a number needs a caveat to be honest — and the recall completion numbers badly do — the caveat sits next to the number, not in a footnote.

Published studies

2026-11-04 The recall repairs that never happened 42,518,965 vehicles carrying a repair nobody did, across 772 campaigns whose reporting window has closed.2026-11-04 Recalled, with no fix and no date 236 campaigns were filed with no repair available. 199 never say when one is coming.2026-11-04 When owners first complain: the odometer, by nameplate 39,830 NHTSA complaints with a mileage, 2018–2022 model years: the typical problem is reported at 23,900 miles, and 79 nameplates are ranked against vehicles their own age.2026-11-04 Same trim, next model year: the makers' own price moves 394 matched trims on 60 nameplates: the 2026 launch price was +1.8% over the 2025 price for the same trim at the median; 3.3% were left unchanged.2026-11-04 Which systems go first: the odometer, by component 58,168 complaint readings, 2018–2022 model years: electronic stability control complaints arrive earliest and engine latest, with the nameplates that report each system earliest.

About these studies

Source: NHTSA public records · manufacturer press kits

Who are these for?

Anyone who wants to know what actually happens to a car before they spend money on one. If you own one of these models, a study tells you what tends to fail, at what mileage, and whether the manufacturer has ever paid for it. If you are shopping, it tells you which model years to walk away from. If you work on cars, write about them, or argue about them for a living, every figure links back to the record it came from and the whole table is attached as a file you can check.

Where do the numbers come from?

Two places, both public. The federal record — NHTSA's recall, complaint, investigation, early-warning and crash-test files — and the manufacturers' own press kits and service bulletins. Nothing here is scraped from dealer listings or submitted by users, and no figure is an estimate unless the page says so.

Does a complaint mean the car is defective?

No, and we never say it does. A complaint is one owner's account, filed with the government and not verified by it. What makes a complaint useful is the pattern: hundreds of owners of the same model year describing the same failure at about the same mileage. Where a study counts complaints, it says so, and it gives you the rows so you can read them yourself.

Can I republish the charts and tables?

Yes, with a credit to WhichTrim and a link. The underlying records are public-domain federal data or the manufacturers' published press kits; the joins and the arithmetic are ours.

How often do these update?

Each study is regenerated with the site, weekly, from the current federal files and press kits. The figures on a study page are as of the date it carries, and they move as the files move.