The least reliable cars on record

Why this is a proportion and not a count

The obvious way to build this page is to add up fires, crashes and deaths and sort by the total. We did that first, and it produced a ranking of America’s best-selling vehicles. The Ford F-150 came first, six more Fords filled out the top ten, and the reason was arithmetic rather than engineering: the more of something there is on the road, the more of everything gets reported about it, including the serious things.

So the ordering is a ratio instead. Fleet size inflates the number of severe complaints and the number of complaints overall in the same proportion, so dividing one by the other cancels it out. What survives is a different and more useful question: when someone does complain about this car, how likely is it that they are describing a fire rather than a rattle.

A worked example

The 2007 Toyota Camry is the clearest case. It has 2,888 complaints on file — more than almost anything else of its era — and under the count-based ordering it sat at number 10 on this page. By share of serious failures it sits at 649. Nothing about the car changed; the arithmetic stopped rewarding it for being common.

The detail is in what those complaints are about. Only 2.2% of them mention a fire and 9.4% a crash. The three largest clusters are engine (338 complaints, 1 crash), vehicle speed control (320 complaints, 93 crashes) and visibility (279 complaints, 0 crashes). These are real problems and owners were right to file them. They are mostly problems of durability and irritation, not of danger, and a ranking that cannot tell those apart is not measuring what its heading claims.

For the same reason, the 1994 model year of the same car carries a severity share of 4.46 against this one’s 1.05 — several times higher — and never appeared on the old list at all, because far fewer people filed anything about it.

What the proportion still cannot do

It is not incidents per thousand vehicles. That would need production or registration figures, and NHTSA publishes neither. The one field that looks like a fleet size — the number of vehicles a recall potentially affected — is the sum of every campaign’s scope, so it reaches 22.9 million for a single model year of F-150 and counts the same truck once per recall. It cannot be a denominator.

You will also see the same model appear several years running near the top. That is not a fault in the ordering: a defect usually spans a generation, so when one year of a car is bad on this measure its neighbours generally are too.

What this page does not cover

There is no ranking of the most reliable cars here, because the data cannot support one. NHTSA collects complaints; a model with few of them may be genuinely sound, or rare, or owned by people who never file. Absence of evidence is not evidence of quality, and a "most reliable" list built from this source would be inventing a signal that is not in it.

Recalls that carry a do-not-drive advisory → · Every manufacturer on file → · How the figures are calculated →