Methodology and data sources
Every number on this site comes from a government dataset. This page says which one, what we did to it, and what we deliberately do not publish.
Data sources
| What you see | Where it comes from | How often it updates |
|---|---|---|
| MOT history, advisories, defects, mileage, expiry | DVSA's MOT history records, the same record behind the official checker | Live: what DVSA holds at the moment you check |
| MOT test station listings | DVSA active MOT test stations register, published on GOV.UK | Quarterly, when DVSA republishes the register |
| Distance sorting and "near me" | Each station's postcode, geocoded against the ONS Postcode Directory | Quarterly |
| Pass rates, common failures and mileage bands by model | DVSA anonymised MOT tests and results, 2023 to 2025 | Annually, when DVSA publishes the previous year |
| Local failure rates by area | The same anonymised dataset, grouped by the postcode area recorded on each test | Annually |
| Car tax rates and bands | Published Vehicle Excise Duty rate tables | Each April, or when rates change |
All of it is published under the Open Government Licence v3.0, which permits commercial reuse with attribution. Attribution appears in the footer of every page.
How the reliability figures are built
The anonymised dataset contains every MOT test carried out in Great Britain, with the make, model, first-use date, mileage, result and each defect recorded. It is published as raw files several gigabytes in size. We process those files offline, once per release, into summary figures. Nothing is estimated or modelled: every percentage is a count divided by a count.
- Which tests count. Normal tests only, class 4 vehicles only, which means cars, taxis and light vans up to 3,000kg. Retests, abandoned tests and other classes are excluded, so the pass rate reflects a car turning up for its annual test.
- What counts as a failure. A test recorded as failed, or as passed after rectification at the test station. In the second case the car arrived with a defect that would have failed it, so we count it as an initial failure. This matches the approach used in the Department for Transport's published summary statistics.
- Model year. The year the vehicle was first used, which is what a buyer sees on the V5C, rather than the year of manufacture.
- Model names. DVSA records the full variant string, for example "A 200 AMG LINE EXECUTIVE". We shorten it to the model designation so that all trims of a model are grouped together, and merge obvious duplicates such as "RAV4" and "RAV-4". The same rule is applied to the live lookup, which is how your car is matched to its model average.
- Minimum sample. A model-year is published only when it has at least 1,000 tests behind it. A postcode area is published only above 5,000 tests. Below those thresholds the figures are not reliable enough to show.
- Common failures. For each model-year, the share of failed tests that recorded each defect. Percentages can add to more than 100 because one test can record several defects.
- Advisories. The share of all tests on which each advisory was recorded, and, nationally, how often a car given an advisory in a category failed on that category at its next test. Each car is followed from one test to the next through its anonymised vehicle identifier.
- Mileage bands and age. Pass rate is also broken down by the mileage recorded at the test and by the vehicle's age at the test, so that a comparison can be made against cars in similar condition, not just the same model.
- Age-adjusted comparison. Because older cars fail more, a model's pass rate at each age is compared with the national pass rate at the same age. The difference, weighted by tests, is the age-adjusted figure shown on reliability pages. Zero is average; positive is better than a typical car of the same age.
- Local failure rates. Each test carries the postcode area of the test station. We compare each area's failure rate and its most common defects against the national figure, and show the difference.
What we do not publish, and why
- Pass rates for individual garages. The public dataset is anonymised and contains no station identifier, so per-garage figures cannot be calculated from it. Any site showing them is estimating.
- Live tax status. DVLA does not currently make live tax status available to third parties. We show the annual rate that applies to a vehicle from the published tables and link to the official checker for whether it is currently taxed.
- Finance, write-off, stolen and keeper history. These are held by insurers, finance houses and the police, not in open data. We do not sell a report containing them.
- Anything about the person who owns a vehicle. The data we use contains no personal information and neither does this site.
Known limitations
- MOT history begins in 2005, when the system was computerised. Older tests are not recorded anywhere.
- A vehicle's first MOT is due at three years old, so there is no history for cars younger than that.
- Mileage readings are as entered by the tester. An obvious typo will show as an anomaly, and we flag it rather than hide it.
- The reliability figures describe how a model performs across the whole fleet. They cannot say how one particular car has been treated.
- Northern Ireland runs a separate MOT system. The station register and the anonymised dataset cover England, Scotland and Wales.
Corrections
If a figure looks wrong, tell us through the contact page. We check every report against the source data and publish a correction where one is due.