Compare two model years
Two used cars, the same page: what owners reported on each, at what mileage, how often per thousand built, and the risk score for both. Give the model and the year for each side.
Pick from the suggestions if you can — that sends the exact vehicle. Typed names are matched against NHTSA's spelling, and where a name is ambiguous we take the model with the most owner reports and say so on the result page.
Comparing across model years is fair; comparing across decades is not. Each risk score is a rank inside its own model year, so a 2009 scoring 30 and a 2019 scoring 30 were each about equally troublesome relative to what else was on sale. What the score cannot tell you is that the 2009 has had ten more years to rust.
What the comparison actually shows
Four things, in the order they appear. The two summary plates hold the figures that describe each car as a whole: its risk score out of 100, the number of owner reports behind it, the mileage by which half the dated reports had been filed, reports per 1,000 vehicles built where a fleet estimate exists, the count of recall campaigns, and the failure mode reported most often. Then the two scores and the two rates are drawn as bars, because a gap is easier to judge by length than by arithmetic.
Underneath that is the part most people come for: the failure modes reported on both cars, ranked by combined report count, with each car's median mileage for that failure side by side. A row is only marked as earlier on one of the two when the medians are more than 10,000 miles apart, because a smaller gap drawn from a few dozen readings is noise wearing a number's clothing. Last come the failures reported on one car but not the other, which are listed rather than scored, because absence of complaints is weak evidence in both directions.
The verdict sentence at the top follows the same restraint. It refuses to name a better car when the two scores are within five points of each other, which happens often and is the honest answer when it does. Two medians a thousand miles apart, drawn from samples of forty, are the same number, and a tool that declares a winner on every comparison is a tool that will eventually be confidently wrong about the car you buy.
Good comparisons to run: two consecutive years of the same model, to see whether a known fault was fixed; the same year of two rivals you are choosing between; or the year you already own against the year you are thinking of upgrading to. A comparison that is rarely worth much is a small car against a large one, or a new model year against one from the far end of the range — the numbers will differ, but for reasons that have nothing to do with how either was built.
Questions this page answers
How do I compare two used cars on reliability?
Give the model and the year for each side above. The result page puts the two risk scores next to each other, lists every failure mode reported on both with the median mileage each was reported at, and names which of the two tends to suffer it earlier. All of it comes from owner complaints filed with NHTSA, so it describes what owners reported rather than what a reviewer thought of the car when it was new.
Can I compare different model years of the same car?
Yes, and it is one of the more useful things this tool does. A mid-generation revision often fixes the failure the earlier years are known for, and putting a 2013 next to a 2016 of the same model shows whether it did. Because each score is a rank inside its own model year, the two remain comparable even when the years are not adjacent.
Does a lower risk score mean the better buy?
Not by itself. A gap of fewer than five points is not a result at these sample sizes, the score says nothing about age, rust or how either example was treated, and it cannot see purchase price. Use it to decide which of the two deserves the closer inspection, then let the inspection and the service history decide.
Source Owner complaints and recall records published by the US National Highway Traffic Safety Administration, model years 1996–2027 Figures rebuilt Published by AtWhatMileage How every figure is computed