About Fault File

The public record of what breaks on British cars, turned into something you can read before you buy one.

Every car in Britain over three years old takes an MOT each year, and the DVSA publishes the anonymised results: which cars passed, which failed, and the recorded reason behind every failure. That runs to more than 40 million tests a year, going back to 2005. Almost nobody turns it into something a car buyer can actually use.

Fault File does. Pick a model and you get its record: how often it fails its first MOT at each age and mileage, and the faults testers write up most, set against the average for every model we track. For a 2018 Audi A3 you can see that the front springs and dampers are its one real weak spot before you ever go and look at one.

Two kinds of known problem

There are two ways to know what goes wrong with a car, and they overlap less than you would expect.

One is what the data proves. The MOT records real faults on real cars in huge numbers, so when the Nissan Qashqai shows corroded brake pipes on one test in six against a class average of one in seventy, that is not hearsay. It is a pattern with sixty thousand cars behind it. That goes in the Known section, and every figure there comes from counted tests.

The other is folklore. The mechanic on YouTube saying "they all do that" is usually right, but the fault being described, a clutch that goes at 70k or an infotainment screen that dies in the cold, tends never to reach an MOT. The test checks that a car is safe and legal to drive. It says nothing about the thing that is about to cost you a grand. That knowledge lives on owner forums and in the trade, so we gather it, put it in the Unconfirmed section, and tag each item as a confirmed pattern, emerging, or disputed. When the measured data begins to bear one of these out, it graduates across.

We keep the two apart on purpose. Plenty of sites fold them into one confident paragraph. We would rather label the proof as proof and the folklore as folklore, and let you weigh them yourself.

Wear tells you about the owner

Roughly a quarter of every model's advisories are worn tyres, and that is a fact about how the last owner drove rather than anything built into the car. So the report splits the two. Faults that judge the model, springs, suspension, oil leaks, a warning light that will not go out, sit in one group. Wear and upkeep, tyres, brake pads, a chipped windscreen, sit in another, greyed back. Both are on the page. Only one of them says anything about the car in front of you.

Why this did not already exist

The data has been free for years. Honest John built a version of this around 2017, and it was good, then the updates stopped. It now serves figures from before the MOT test itself was overhauled in 2018, and every test since has sat unused. That seemed like a waste.

Made by appy lad

Fault File is built by one person under the name appy lad, alongside a few other small things I make. It is independent, with no connection to the DVSA or to any manufacturer, dealer or warranty firm.

Found a mistake, or know a fault the data has not caught yet? Tell me.
hello@appylad.com

What it cannot tell you

Some honest limits. The MOT only sees roadworthiness, so engines, gearboxes and clutches barely register in the Known section. Models are split into generations by first-registration year, and a car registered in a changeover year could sit on either side of the line. And a low failure rate on a fifteen-year-old car can simply mean the rough ones were scrapped and the survivors are the ones being looked after. Read the mileage view alongside the age view, and treat the whole thing as a strong steer rather than a promise about one particular car.