High-Mileage Cars to Avoid, and How We Worked It Out

The method matters more than the list, because the list is only as good as the thing being counted, and the thing usually being counted is wrong. So the method is set out first, and every number is one you can recount yourself.

The list everyone publishes is a popularity contest

The raw material for almost every “cars to avoid” article is the federal complaint database. It's free, it's public, and it's enormous. It's also a trap, and the trap catches nearly everybody.

The Ford Focus has 7,631 complaints across the three model years we scanned. The Toyota RAV4 has 696. Read that as a reliability verdict and you've concluded that Ford builds cars eleven times worse than Toyota. What you've measured is that Ford sold a great many more Focuses.

NHTSA publishes no sales figures, so there's nothing to divide by. Every ranking of the most unreliable used cars built on complaint counts is a list of popular cars wearing a warning label, and the sites publishing them rarely say so.

A measure that survives the sales figures

There's a way around it, and it's almost embarrassingly simple: stop comparing models to each other and compare a model to itself.

Of everything the owners of one car complained about, what share described a failure that ends in a four-figure bill. A failed engine, a dead transmission, a stretched timing chain, a head gasket, oil consumption serious enough to write in about? Sales volume inflates the top and the bottom of that fraction equally, so it cancels out. A model that sold five million units and one that sold fifty thousand can be set side by side honestly.

We ran it over 22 mainstream models at the 2012, 2014 and 2016 model years, cars that are high-mileage used buys today, and kept only models with at least 300 complaints, below which the share is noise. The spread isn't subtle. The lowest is 2.1 percent. The highest is 25.9.

One thing to hold on to before reading the order: this isn't a probability. It doesn't say the Equinox is twelve times more likely to break than the Corolla. It says that when an Equinox owner writes to the regulator, the odds are roughly one in four that they're describing something that will cost four figures, and for a Corolla owner, roughly one in fifty.

The high mileage cars to avoid, and what each one does

A share on its own is an accusation without evidence. The same scan says what the complaints are about, and each of these models has one dominant story rather than a scattering of unrelated faults.

That concentration is what makes the ratio worth reading. These aren't cars with a little of everything wrong. They're cars with one specific, well-known, expensive thing wrong, reported over and over by different owners in their own words.

The Ford Focus is the one that nearly got away. Our first pass looked for the words owners of a broken transmission are assumed to use, failed, slipping, replaced, and it put the Focus at 4.5 percent, apparently safer than a CR-V.

Anyone who lived through the DPS6 dual-clutch would have known instantly that this was wrong. Those owners don't write “transmission failure.” They write that it shudders, jerks and hesitates, and 629 of them did. Widening the filter to the words people use moved the Focus from 4.5 percent to 15.4, and barely moved anything else.

The ones the same test puts at the bottom

The bottom of the table is worth as much as the top, and it's the part a buyer can act on most cheaply. Toyota Corolla, 2.1 percent. Jeep Grand Cherokee, 2.4. Chevrolet Malibu, 2.6. Honda Civic, 3.2. Toyota RAV4, 3.4. Honda Accord, 3.8. Ford Fusion, 3.9.

Some of that is exactly what you would guess, and it's worth noticing that a measure which produced a surprising order everywhere would more likely be broken than insightful. A metric that puts the Corolla at the safe end and the Nissan CVT cars at the other is at least measuring something real.

Where this measure lies to you

Now the part that matters, because every method has a failure mode and most articles don't go looking for their own.

The denominator is every complaint about the car, whatever it is about. So a model with one enormous non-drivetrain problem gets a flattering score for a reason that has nothing to do with its drivetrain.

The Toyota Corolla is the clearest case. It scores best in the whole table at 2.1 percent, and 54 percent of its complaints are about Takata airbag inflators. That recall swelled the bottom of the fraction enormously, and the drivetrain share fell as a matter of arithmetic. The Corolla is a genuinely reliable car, but this particular number is flattering it, and we would rather say so than let a good result stand on a bad foundation.

The Jeep Grand Cherokee is the same shape of problem. It looks safe at 2.4 percent, on 4,318 complaints of which 18 percent are electrical and 17 percent brakes. Its owners had a great deal to complain about. Very little of it was the drivetrain, which is what this measure asks about and all it can answer.

There are three further limits, and they apply to any use of this database.

• Filing a complaint tracks anger, warranty status and press coverage as much as it tracks metal. A class action or a recall drives filings on its own.
• The database is a safety file, so an expensive fault that never made a car unsafe is under-reported.
• The filters are keyword matches over text people wrote themselves, so they miss complaints that describe a fault without naming it — exactly as ours did with the Focus until we caught it.

How to use this when you're standing in front of a car

None of the above tells you whether the specific car in front of you is sound. It tells you what to check hardest, which is a different and more useful thing.

If the model is near the top of that table, the failure isn't a mystery: it's one named component, and it can be tested on a viewing.

A Nissan with a CVT should be driven from cold and made to accelerate from a steady 40 mph. A Focus should be crept through traffic, because the dual-clutch signature shows at walking pace and nowhere else. A car with an oil consumption reputation needs the dipstick read twice, a hundred miles apart, which means two visits.

And a seller who has warmed the engine before you arrive has quietly removed two of the five tests worth doing. That's not proof of anything, but it's a reason to come back another morning.

What the whole class of these cars costs to keep once you've bought one is a separate question, and a bigger one: see the 100k mile maintenance cost breakdown. If it is oil use specifically that worries you, the thresholds are stranger than they look — every published figure for normal consumption expires at 36,000 miles.

Common questions

Does a high complaint share mean the car will break?

No, and this is the misreading to avoid. It means that among owners who complained, a large share complained about something expensive. A car can have a high share and a low failure rate if few owners have trouble at all. It narrows what to inspect; it doesn't predict your car.

Why only 2012, 2014 and 2016?

Those are old enough to be genuinely high-mileage purchases now and recent enough that the complaint file is well populated. Sampling every year would have tripled the scan without changing the shape of the answer.

Should I avoid every car on the list?

No. A Nissan with documented CVT replacement is a better buy than an untouched one, and the same is true of a Focus with the clutch pack done. The list tells you which question to ask the seller, and the answer can be reassuring.

Where do used cars with engine problems show up here?

In the engine-failure column, mostly. The Kia Optima and Hyundai Sonata are the clearest cases in this scan: 224 and 267 complaints respectively say the engine failed outright, with oil consumption behind it. That's a different failure from the transmission cars above it, and a different conversation with a seller.

What about brands not on the list?

They weren't scanned, and absence isn't a clean bill of health. We picked 22 mainstream models with enough complaints to be meaningful. A model missing here is missing, not cleared.

Written for the United States and Europe, compiled in August 2026. This article is general information, not advice about your specific vehicle, see the disclaimer.