Home How the scores are made CarsThat Last
EN

How the scores are made

Where the data comes from

Two countries publish inspection results car by car. The Dutch vehicle authority RDW publishes every defect found at the APK inspection. The UK's DVSA publishes every MOT test with its defects and mileage. Together that is tens of millions of cars.

For each car we use its most recent inspection. Campers, large vans, pickups and other special vehicles are left out where the data lets us tell.

What counts as a defect

We score technical defects in six areas: brakes, suspension and steering, engine, drivetrain and leaks, exhaust and emissions, rust and structure, and warning lights, airbags and belts.

Tyres, bulbs, wipers, mirrors and number plates are counted but not scored. They wear out on every car and mostly show how well the owner looks after it. Advisory notes are not scored either.

A fair comparison

Old cars have more defects than young ones, and diesels are driven further than petrol cars. So each model is compared with the average car of the same age and fuel in its own country. In the UK we also match on mileage.

Hybrids and plug-in hybrids count as petrol cars for this comparison, because that is the choice a buyer makes. Diesel, electric and LPG cars are each compared with their own kind.

The result is an index. At 1.0 a model has exactly the defects you would expect. At 0.5 it has half, at 2.0 double.

The Dutch data has no mileage. Models are driven in much the same way in both countries, so we apply each model's UK mileage correction to its Dutch result. We tested this: it brings the two countries closer together.

Combining countries

Each country is first measured against its own average, which cancels out one country testing more strictly than the other. Then the results are combined. A country with plenty of cars for a model counts about as much as the other, however large its fleet. A country with only a few hundred counts less.

Countries often disagree about a model, and electric cars most of all. When they do, the likely range gets wider. Models with few cars are pulled towards the typical model, so a lucky run of clean inspections cannot put a rare car on top.

From index to score

The score runs from 0 to 10: score = 10 / (1 + 2/3 × index2.5). An average car scores 6.0. Half the usual defects gives about 8.9 and double gives about 2.1. From 8.5 we call a score excellent, from 7 good, from 5 average, from 3.5 below average and anything lower poor.

The verdict

Next to the score is a verdict against cars of the same type, because a sports car and a van are not driven alike. "Better than most" means the whole likely range is better than the typical car of that type. "Probably better" means it is at least 80% likely. The same steps exist on the worse side. Everything in between is about average.

Years, fuels and areas

Year scores show how cars first registered in that year do, relative to the model as a whole. Fuel scores use the same comparison as the model: hybrids against petrol-engined cars, the others against their own fuel. The UK data cannot tell plug-in hybrids from other hybrids, so the two are scored together. Area scores and the faults under "What to check" use the same comparison per defect.

What the score cannot tell you

Inspections find wear, leaks and corrosion. They do not find most engine, gearbox, battery or electronics failures, and they never see a fault that was repaired the week before. Scrapped cars drop out of the data, which flatters old cars a little. Cars under about three years old are not inspected yet, so new models appear late.

A score describes a model, not the car in front of you. Have a used car checked before you buy it.

Sources and licences

RDW open data, licensed CC0. DVSA anonymised MOT tests and results, which contains public sector information licensed under the Open Government Licence v3.0. Scores, rankings and text on this site are our own work.