How we measure
Every number on this site comes from the raw inspection records. This is exactly how.
Source
DVSA, Anonymised MOT tests and results, 2024 test year, published under the Open Government Licence v3.0. Two files: one row per test (make, model, fuel, result, mileage, first-use date) and one row per defect item, joined on the test identifier. The defect wording comes from the DVSA item-detail lookup table.
Population
442,243 first-time tests of class 4 vehicles recorded as electric. Only models with at least 100 tests are published; that leaves 83 models covering about 99% of the tests.
Identifying electric cars
Fuel type recorded as electric under either of the two codes the file uses, and engine capacity zero or blank. The second condition matters: several hybrids are recorded as electric and would otherwise inflate the failure rate. A short manual exclusion list handles plug-in hybrids that slip through with a blank engine field.
Failure rate
Tests with result "fail" divided by all first-time tests. Retests are excluded, so a car that fails and returns counts once. A pass after rectification at the station counts as a pass. Rates by age are hidden where fewer than 50 cars of that age were tested.
Defect shares
Every defect on a failed test is assigned to the top-level inspection group it belongs to (tyres, suspension, brakes and so on). A model's share for a system is that system's defects divided by all its failing defects. Shares describe what a car fails on, not how often it fails; the two are shown separately on purpose.
Comparison group
Petrol and diesel cars from the same test year, aged three to eight years. The failure rate by age uses the full population; the defect breakdown uses a 10% sample of failed tests.
What is not measured
Battery health, motor and charging system. The inspection does not test them. Owner reports are collected separately, summarised in our own words, linked to their source and never counted.
Data quality notes
The published files contain duplicated rows and free-text model names that include trim levels. Rows are de-duplicated and models normalised to make plus base name; the trim strings are shown on each model page as "versions seen in the tests". Fuel codes were found to be inconsistent within a single file; see above.
Updates
Recalculated when DVSA publishes a new test year. Last refresh 2026-09-13.