By the only measure that is public and verifiable — complaints filed with NHTSA per production run — the cleanest mainstream marques are BMW, GMC, Subaru. The heaviest is Nissan at 5.7 per run against 2.34 for the best, a 2.4-fold spread across 23 brands with enough records to compare. But the badge is not where reliability is decided: half of all complaints arrive by 5 years of age while basic warranties run three, so 73.1% of failures land after the manufacturer stops paying — which makes the year you buy matter more than the maker.
When cars start to break
Not the calendar year the complaint was filed — the age of the car when it failed. That age decides who pays: inside the warranty the manufacturer does, outside it you do.
| Age at failure | Complaints | Running total | Warranty |
|---|---|---|---|
| 0 years | 9.7% | covered | |
| 1 year | 18.5% | covered | |
| 2 years | 26.9% | covered | |
| 3 years | 36.7% | you pay | |
| 4 years | 47.1% | you pay | |
| 5 years | 57.5% | you pay | |
| 6 years | 67% | you pay | |
| 7 years | 75% | you pay | |
| 8 years | 81.7% | you pay | |
| 9 years | 87.1% | you pay | |
| 10 years | 91.2% | you pay | |
| 11 years | 94.3% | you pay | |
| 12 years | 96.5% | you pay | |
| 13 years | 98.1% | you pay | |
| 14 years | 99.2% | you pay | |
| 15 years | 100% | you pay |
The market pattern: failures peak at four to five years, while basic warranties stop at three. That is why a three-year-old car off lease looks like a bargain and a five-year-old one is where the bills start.
Reliability by brand
Complaints per production run across every model of the marque. Fewer is better. Note that this measures how often owners complained to the federal regulator — a serious bar, quite different from a satisfaction survey.
| Brand | Per run | Complaints | Models | Crash-related |
|---|---|---|---|---|
| Lamborghini | 9 | 8 | 0% | |
| Peugeot | 9 | 2 | 0% | |
| Ferrari | 43 | 71 | 9.3% | |
| Renault | 34 | 3 | 8.8% | |
| Aston Martin | 13 | 35 | 0% | |
| Daihatsu | 10 | 2 | 10% | |
| Bentley | 7 | 9 | 28.6% | |
| Rolls-Royce | 13 | 14 | 0% | |
| Lexus | 295 | 17 | 8.5% | |
| McLaren | 15 | 10 | 0% | |
| Maserati | 161 | 13 | 11.2% | |
| Porsche | 1,390 | 15 | 3.7% | |
| Lucid | 68 | 2 | 20.6% | |
| Saab | 2,182 | 7 | 3.2% | |
| Land Rover | 2,019 | 17 | 4.8% | |
| Jaguar | 1,988 | 17 | 7.8% | |
| Lotus | 97 | 13 | 1% | |
| Audi | 9,238 | 50 | 3.4% | |
| Suzuki | 3,079 | 17 | 9.2% | |
| Mitsubishi | 8,144 | 26 | 11% | |
| Alfa Romeo | 363 | 19 | 7.4% | |
| Genesis | 823 | 6 | 7.2% | |
| Volvo | 5,510 | 39 | 4.7% | |
| Rivian | 261 | 4 | 11.1% | |
| BMW | 13,111 | 105 | 4.4% | |
| GMC | 29,598 | 25 | 5.2% | |
| Isuzu | 2,639 | 27 | 9.6% | |
| Subaru | 22,483 | 21 | 6% | |
| Infiniti | 4,590 | 33 | 6% | |
| Cadillac | 12,964 | 32 | 6.5% | |
| Mazda | 17,028 | 27 | 6.6% | |
| Daewoo | 912 | 3 | 3.3% | |
| Volkswagen | 25,925 | 34 | 3.8% | |
| Acura | 11,238 | 17 | 5.2% | |
| RAM | 16,974 | 9 | 4.3% | |
| Lincoln | 11,420 | 17 | 5.9% | |
| Buick | 16,826 | 25 | 5.4% | |
| Fiat | 1,200 | 7 | 4.3% | |
| Mercedes-Benz | 16,003 | 40 | 4.1% | |
| Dodge | 70,321 | 43 | 6.1% | |
| Smart | 250 | 2 | 4.4% | |
| Honda | 77,453 | 21 | 6.5% | |
| Chevrolet | 123,800 | 62 | 5.8% | |
| Scion | 1,877 | 8 | 12.7% | |
| Toyota | 87,324 | 43 | 10% | |
| Saturn | 10,411 | 14 | 5.6% | |
| Pontiac | 19,042 | 17 | 5.5% | |
| Ford | 207,951 | 59 | 4.8% | |
| Chrysler | 29,734 | 36 | 5% | |
| Hyundai | 47,965 | 35 | 5.4% | |
| Mini | 3,426 | 8 | 2.4% | |
| Kia | 38,051 | 25 | 5.3% | |
| Jeep | 73,283 | 22 | 5.1% | |
| Nissan | 58,260 | 36 | 5.1% | |
| Polestar | 637 | 4 | 1.3% | |
| Tesla | 6,071 | 6 | 10.8% | |
| Fisker | 860 | 4 | 1.4% |
Why this list looks nothing like the surveys
Open any «most reliable brands» ranking and Toyota is at the top. Open ours and it sits 15th of 23. We are not claiming the surveys are wrong — we are measuring something else, and the difference is worth understanding before you trust either.
A complaint is not a survey answer
Consumer Reports and J.D. Power ask owners what annoyed them, and everything counts: a rattle, an infotainment freeze, a seat that squeaks. We count reports filed with a federal safety regulator, which takes effort and is normally reserved for something that failed. Those are different questions, so different answers are the expected outcome, not an error in either.
Volume distorts the rate — and we can show by how much
Our figure is complaints divided by production runs, not by cars sold, because nobody publishes per-model sales. A mass-market marque packs more vehicles into one run than a low-volume one, so the same fault rate produces a higher number. Measured across the 23 brands here, the correlation between a marque's record count and its complaint rate is 0.58 — strong enough that the biggest sellers are pushed down the table by arithmetic alone, which is most of the reason Honda (3.69 per run) and Toyota (4.03) rank where they do.
Silence is not excellence
Brands with almost no US owners collect almost no complaints and would top any raw ranking — Lamborghini and Peugeot sit at nine records each. That is an absence of drivers, not an absence of faults, so this page drops every marque under 10,000 records. 34 of them fall away, and what remains is comparable.
What to take from it
Use the table above for what it honestly is — how often owners of a marque escalate to the regulator — and use the age curve for the decision that actually costs money. Half of everything reported arrives around 5 years, past the basic warranty on most cars, and that is true whichever badge wins the survey this year.
What this means when buying used
A used car is bought at a specific age, and that age decides who pays for the next failure. The curve above puts the median complaint at 5 years, while basic cover on most makes runs three — so the question is not whether the brand is reliable in general, but where on that curve the particular car sits.
| Age at purchase | Where it sits on the curve | What it means for you |
|---|---|---|
| Under 3 years | Inside basic cover on most makes | Roughly 26.9% of reported failures happen here — the maker pays for them |
| 3 to 5 years | Past basic cover, before the median failure | The cheapest window to own: warranty gone, but the bulk of faults not yet due |
| 5 to 8 years | Around and after the median complaint | Failures cluster here and the bill is yours — budget for it or buy cover |
| Over 8 years | Long tail of the curve | Fewer new reports, but survivors have wear that no complaint database records |
Two checks are worth more than any brand ranking at this point. Run the exact VIN — build runs of the same model differ from each other far more than brands differ from brands. And look at the model year rather than the marque: on most models the gap between the best and worst year is wider than the gap between the best and worst brand in the table above.
Good year or bad year — the VIN settles it.
- Check any VIN against complaints for its exact build year and plant
- Exact value for that car, adjusted for its mileage
- Full research report as a PDF you can hand to a mechanic
- Alerts if a recall lands on a car in your garage
- No ads, on every page
- No ads, no tracking — and much more
How we measure it
Reliability scores usually arrive as a number out of five with no working shown. We publish the arithmetic instead, because two things about it are arguable and you should be able to argue with them.
Rate, not total. Volume alone would decide the ranking, so we divide by the number of production runs — the distinct first-eleven-character VIN prefixes on record. Across 1,100,126 complaints and 295,368 runs the market average works out at 3.72 per run.
Age, not year. Incident date minus model year gives the age at failure, which is what the warranty question turns on. It is why our «worst year» rankings exclude anything built in the last 6 years — those cars simply have not had time to break yet.
What this is not. Filing with a federal regulator takes effort, so what you see are faults someone judged serious — not scratched paint, not running costs, not how nice the car is to live with. And an absence of complaints on a rare model means an absence of owners, not excellence.
Frequently asked questions
How is car reliability measured here?
Complaints filed with NHTSA, divided by the number of production runs on record. That turns a raw count into a rate, so a best-seller is not penalised for selling well.
At what age do cars start breaking?
Half of all complaints arrive by 5 years, with the peak at four to five. Basic warranties end at three, so most failures land after the manufacturer stops paying.
Which car brand is most reliable?
By complaint rate, Lamborghini at 1 per production run. The table above ranks every brand with enough records.
Is this the same as Consumer Reports?
No. Those are owner surveys — opinions, gathered from subscribers. This is federal complaint records — actions, taken by people who had a fault serious enough to report to a regulator. The two measure different things and often disagree.