Research

UK MOT results 2025: which used cars fail, and why

In 2025, 27.7% of the 33,210,823 class 4 first MOT tests in Great Britain ended in an initial fail: a fail, or a fail that was fixed at the test station. That is DVSA’s own measure, and it is the one used throughout this page. Here is how it changes with the model, the age and mileage of the car and the area it was tested in, and which parts cause the failures, from the complete 2025 DVSA file.

MOT results only: the test does not look at the engine, clutch or gearbox. Analysis by Autoza of DVSA’s MOT anonymised data for 2025. Last updated 10 November 2026.

Key findings

1. At the first MOT, small cars differ.A Vauxhall Corsa failed its first MOT 14.3% of the time (29,897 tests) and a Ford Fiesta 9.1% (21,859 tests): 57% more often. The Hyundai i10 (16.4%, 7,783 tests) and i20 (15.6%, 6,504 tests) failed more often than the Corsa, so this is a spread across small cars, not one model.
2. At ten years old the spread is wider.At ten years old a Renault Clio failed 44.5% of the time (24,050 tests) and a Volkswagen Golf 18.1% (67,353 tests), about 2.5 times as often. The first-MOT gap between the Corsa and the Fiesta has closed by then: 36.7% and 35.2%.
3. Mileage is a second clock, up to a point.For cars aged 6 to 7, the initial fail rate was 9.2% under 20,000 miles (350,188 tests) and 17.9% at 40,000 to 60,000 miles (1,491,480 tests): roughly double. It reaches 22.2% at 80,000 to 100,000 miles and 22.4% at 100,000 to 120,000, then eases to 19.3% at 160,000 miles or more, probably because only the better-kept cars reach those mileages.
4. Cars that barely move fail about half as often.Cars aged 5 to 14 driven under 3,000 miles a year failed 13.4% of the time, against 24.7% for cars of the same age (0.54 times; 1,299,849 tests). Annual mileage here is the odometer divided by age, and low-use cars may also be better looked after, so this is not proof that low mileage itself prevents failures.
5. Where a fail is fixed on the spot depends on the area.45.6% of initial fails at test stations in East London (268,239 tests) were fixed and passed at the station, against 7.9% in Aberdeen (254,259 tests) and 19.0% across Great Britain. That says how test stations work, such as whether they repair on site, not how anyone drives.
6. The part behind the failure often depends on the model.Against cars of the same age, ball-joint dust covers appeared on 6.7% of Vauxhall Mokka tests (19 times the rate), track rod end play on 5.0% of Nissan Qashqai tests (5.2 times), a broken coil spring on 8.7% of Renault Clio tests (3.2 times) and a split CV boot on 6.1% of Peugeot 2008 tests (8.4 times). Most are routine repairs; a broken coil spring or a worn ball joint is the kind of safety item the test exists to catch.
7. Tyres stand out on the Tesla Model 3 and Model Y.Tyre defects appeared on 11.2% of the 87,481 Tesla Model 3 tests and 11.7% of the 34,928 Model Y tests, against 5.9% expected for cars of the same age and mileage: 1.9 times and 2.0 times. Brake defects were well below par, which fits regenerative braking taking much of the work.

Every figure is the DVSA-basis initial fail rate unless it says “outright”. Method, definitions and caveats are below.

How the same model fares as it ages

Failure rises with age for every model, but not at the same speed. These tables show the initial fail rate for popular models at their first MOT, at 4 to 5 years, at 7 to 8 and at 10, with the number of tests and the 95% margin. The last column compares all ages together with cars of the same age and mileage, which takes out the fact that a model sold for years is mostly old cars on the road. Models are listed A to Z within each group: this is not a league table, and the figures describe a model, not any one car.

The small grey line under a first-MOT or ten-year figure is the outright fail rate, where a car fixed at the station counts as a pass. “All ages” is the model’s rate across every age, against the rate for cars of the same age and mileage, with the gap in percentage points. A dash means fewer than 1,000 tests. A model name links to our fault guide for it.

Mileage is a second clock

Cars of the same age fail more as the odometer climbs, up to a point. For cars aged 6 to 7, the rate roughly doubles between the lowest mileage band and 40,000 to 60,000 miles, keeps rising to about 80,000 miles and is level from there to 120,000. Above that the rate eases, which probably reflects the better-kept cars that reach those mileages.

Cars that hardly move

Grouping cars aged 5 to 14 by the miles they cover each year (the odometer reading divided by age) shows the same thing from the other side: the few that cover under 3,000 miles a year fail about half as often as other cars of their age. Annual mileage here is derived, not recorded, and low-use cars may also be better looked after.

Why cars fail, model by model

Every defect on a failing test is filed under one of DVSA’s 13 top-level groups. For each model we compare the share of its tests carrying a defect in the group with the share you would expect for cars of the same age and mileage. The table shows the robust findings only, at most the three largest higher-than-expected and the two largest lower-than-expected for each model; the CSV has every model and every group. A finding is robust where the sample is large and the gap is well beyond chance (the method has the exact rule).

“Share of its tests” is the percentage of all that model’s 2025 tests with at least one defect in the group. A model can be lower than expected in a group and still fail overall, and the other way round.

The parts behind the numbers

The defect file goes down to the specific part. These are the ones behind the headline findings, with the same defect’s rate for cars of the same age band for comparison. This comparison allows for age only, not mileage, and each is a routine repair or a safety item according to DVSA’s grade (Major or Dangerous), not a verdict on the model. A recall is a free fix and is a separate question.

By postcode area: where a fail is fixed on the spot

DVSA records the postcode area of the test station, not of the owner. The table covers all 118 areas with the initial fail rate as tested and allowing for the age of the cars (so an area with an older fleet is not marked down for it), the outright fail rate on the same basis, and the share of initial fails that were fixed and passed at the station. Each area has a page of its own with the figures, how it compares and what fails there.

The share fixed on the spot varies from 7.7% in Kirkcaldy (Fife) to 45.6% in East London, against 19.0% nationally. That reflects how and where testing is done, such as whether a station repairs on site, and not the standard of anyone’s driving. It also changes how big the gaps between areas look: between the areas with the highest and lowest age-adjusted rates the gap is 1.9 times on DVSA’s measure but 3.0 times on outright fails, because so many fails in some areas are fixed and passed on the spot.

Method and definitions

Initial fail rate (the headline everywhere)

The share of first tests that ended in a fail, or in a fail that was repaired at the test station and passed (a “pass after rectification”). This is the figure DVSA publishes. Across Great Britain in 2025 it was 27.7%.

Outright fail rate (shown second)

The share that failed outright, with a car fixed at the station counted as a pass. It was 22.4% nationally. It is lower than the initial fail rate; 100 minus it is the “passed on the day” basis of our model guides’ pass rates, so the two are never mixed in one comparison.

Passed on the day (the figure on our model guides)

A pass, or a fail fixed at the station and then passed: everything that is not an outright fail. Nationally 77.6%. Our model guides’ “MOT pass rate” uses this basis (from a sample of the same 2025 file), so it is about five points higher than 100 minus DVSA’s initial fail rate.

Fixed at the station

A fail repaired and passed at the test station. Across the country 19.0% of initial fails ended that way (1,745,654 tests).

First MOT

A test 1,020 to 1,139 days after the car was first used: the first test at about three years old.

Age bands

Age 4 to 5, 7 to 8 and 10 are the nearest whole year of age at the test. “Cars aged 6 to 7” and “aged 10 to 11” are the same, for the mileage tables.

Same age, or same age and mileage

Wherever a model or area is compared with “expected”, expected is what all cars of the same age (in whole years) and the same 20,000-mile band scored nationally. The model’s own age and mileage mix is therefore taken out of the comparison. The specific-defect table (the parts) allows for age only.

Age-adjusted (areas)

An area’s age-adjusted rate is what its cars would have scored at the national rate for each age, scaled to the national figure (indirect standardisation, age in whole years). The margin is the 95% margin on that figure. A version that also allows for mileage is in the area download.

Failure categories and “robust”

Every defect is filed under one of DVSA’s 13 top-level groups (lamps and electrical, suspension, brakes, tyres and so on). A test counts once per group, however many defects it has in it. A finding is called robust only where there are about 300 or more expected cases, the difference is at least six standard errors (a z-score of 6 or more) and the ratio is at least 1.25 (or at most 0.8). Every model in the table has at least 20,000 tests.

How the data was prepared

• Source: DVSA’s MOT anonymised test results and defect (test item) data for all of 2025, published under the Open Government Licence v3.0. We use every class 4 normal test (cars and light vans up to 3,000kg) that ended in a pass, fail or pass after rectification: 33,210,823 first tests after leaving out retests, aborted tests and a few hundred malformed lines. It is the complete file, not a sample.
• The national figures are close to DVSA’s own published statistics, which cover a different period and classes: for classes 3 and 4 in the financial year 2025 to 2026 DVSA reports an initial failure rate of 27.51% and a final failure rate of 22.30%; for class 4 in the calendar year 2025 we count 27.68% and 22.43% (outright fail, the same idea as DVSA’s final failure). Every failing test in the results file was matched to its defect records.
• One vehicle can count more than once: a car given two normal tests in the year (not a retest) is two tests, and we do not deduplicate. We could not measure how often this happens across the full year (the raw rows were deleted after aggregation); in the sample slices of the same file that we kept, 0.9% of tests came from a vehicle with more than one normal test in them, which understates the full-year share.
• Models: DVSA’s model field is free text typed from the V5C, so we group by make and the first word of the model (with rules for BMW and Mercedes-Benz series, MINI, Tesla and Land Rover), which means a name such as “Citroen C4” includes its variants. Model tables cover cars only: vans and pick-ups were removed by name. A model appears only with at least 20,000 tests in the year, and an age figure is printed only with at least 1,000 tests behind it. Area and national tables are all class 4 first tests, as DVSA counts them.
• Model margins are the usual 95% binomial margin on the printed rate and the number of tests; area margins are computed on the observed fails.
• Areas: DVSA records the postcode area of the test station, not of the car’s owner. A handful of tests recorded against areas too small to report are pooled as “XX” in the area download and not shown on the pages.
• Mileage: tests with an odometer reading of zero were left out. Annual mileage is the odometer reading divided by the car’s age, so it is derived, not recorded.

Definitions version 1.1. Version 1.0 (2026-10-23): First edition: initial fail (fail + fixed during the test) as the headline, outright fail second. Version 1.1 (2026-10-27): No figure changed. Added the three outcomes shown separately (passed with nothing to fix, fixed during the test, failed) and named “passed on the day” (pass + fixed during the test), the basis of the model guides’ pass rates; added fuel as an optional filter in the planner; corrected the comparison with DVSA’s published figures.

The same tests, both ways of counting

A test ends in one of three ways: a pass with nothing to fix, a fail fixed at the station and then passed, or a fail. DVSA’s initial fail rate counts the last two; “passed on the day”, the basis of our model guides’ pass rates, counts the first two. Here are the three outcomes as counted, so either figure can be checked.

The model rows are every age of the model together, so they mostly reflect how old its cars on the road are; the age-matched comparisons above are the fair ones. Which definition is used changes every figure by roughly the share fixed during the test (about 5 points), but it changes almost nothing about how models compare.

Source files, data dictionary and what is proven

The source is DVSA’s anonymised MOT data for 2025, listed on DVSA’s open data page (the data.gov.uk catalogue entry still lists 2023 and 2024 only). Checked 2026-10-27.

• MOT testing data results extracts (2025)4,835,709,507 bytes; DVSA last modified 2026-06-19; DVSA server tag 56919f063c38796f30118731d4794d09-282.
• MOT testing data failure item extracts (2025)4,359,554,510 bytes; DVSA last modified 2026-06-19; DVSA server tag 24717318d6a213f6ebba50ecc98520da-254.

Two cuts of the same file

• The complete 2025 file (all 12 months). January to December 2025 (12 monthly files, 202501 to 202512). 33,210,823 class 4 normal tests. Downloaded from DVSA on 23 Oct 2026 month by month (ranged reads of the two zips above; each monthly member was checked against its size in the zip directory). Raw rows were deleted after aggregation the same day. Used by /research/mot-2025 and its articles, the MOT planner, the fuel-and-age sections of five fault guides, the monthly asking-price list.
• A sample of the 2025 file (about one test in ten, from nine months). January, February, April, May, July, August, September, October and November 2025 (each slice falls within a month or two; March, June and December are absent). 3,459,000 class 4 normal tests. Eight contiguous 60 MB slices of the 2025 results file, downloaded 12 Aug 2026 and kept. Used by the "MOT pass rate" block and the by-year and by-mileage tables on the model fault guides, compare pages, brand pages and the mileage tools.

How closely the sample matches the full year, on the “passed on the day” basis: All class 4 tests 77.3% in the sample, 77.6% in the full year; Nissan Qashqai 73.5% in the sample, 73.5% in the full year; Nissan Juke 73.7% in the sample, 74.6% in the full year; Ford Kuga 83.4% in the sample, 83.7% in the full year; Vauxhall Corsa 71.0% in the sample, 71.7% in the full year; Kia Sportage 80.5% in the sample, 80.7% in the full year.

The sample is not a designed random sample, so this agreement is evidence that it is representative, not proof.

Data dictionary

test_id

DVSA’s id for one test. Joins a test to its defect items. Never published.

vehicle_id

DVSA’s anonymised id for the vehicle. Not used to deduplicate: a vehicle with two normal tests in the year counts twice. Never published.

test_date

Date of the test. Age at the test; the month a test falls in.

test_class_id

MOT class. Class 4 only: cars, taxis, private hire, motor caravans and light goods vehicles up to 3,000 kg.

test_type

NT normal test, RT retest, PL/PV partial retests, ES/EI appeals. NT only. Retests and appeals are left out.

test_result

P passed, F failed, PRS passed after rectification at the station, ABR/ABRVE aborted, ABA abandoned, R refused. P, F and PRS only; the rest are left out (about 0.6% of class 4 normal tests in the sample).

test_mileage

Odometer reading recorded at the test. Used as recorded; the file has no unit column and we convert nothing. A zero or blank reading is “no reading”.

make, model

Free text from the vehicle record. Grouped by make and the first word of the model, with set rules for BMW and Mercedes-Benz series, MINI, Tesla and Land Rover; vans and pick-ups removed by name.

fuel_type

PE petrol, DI diesel, HY hybrid electric, ED electric diesel, EL electric, others. Petrol, diesel, hybrid (HY and ED) and electric; others are kept in all-fuel figures only.

first_use_date

Date the vehicle was first used. Age at the test = days from first use to the test ÷ 365.25, rounded. The first MOT is a test 1,020 to 1,139 days after first use.

postcode_area

Postcode area of the test station. Area tables. It is where the test took place, not where the owner lives.

rfr_id, rfr_type_code (item file)

The defect and its type: F fail, P fixed during the test (PRS), M minor, A advisory; with its grade (Major or Dangerous). F and P items only, filed under DVSA’s 13 top-level groups; a test counts once per group. Minor faults and advisories are not counted.

What is proven, and what is not

• The two 2025 files are DVSA’s: DVSA’s own download page links them by these exact addresses, and their sizes on DVSA’s server equal the sizes our downloader read.
• They hold 2025 tests: the results file has twelve monthly members, 202501 to 202512, and every test date in the slices we kept falls in 2025.
• Every failing test in the full-year results file was matched to defect records (12 of 12 months).
• The committed tables are reproducible from our aggregates: scripts/research/build-mot-products-data.py --check rebuilds them and finds no difference.
• The sample agrees with the full year to within a point for the whole fleet and for the five most-read guide models.

• Not proven: We cannot show a checksum of the raw 2025 files: DVSA publishes none, and our raw copy was deleted after aggregation. We hold DVSA’s size and server tag for each file.
• Not proven: DVSA’s file has no unit column for the odometer reading; we have not confirmed whether any readings were typed in kilometres.
• Not proven: The fuel is what DVSA records. Some of it is plainly wrong (a few Toyota Auris tests are recorded as electric), and hybrids of every kind share one code.
• Not proven: Vans are removed by model name. A van or motor caravan recorded under a car name stays in the car figures.
• Not proven: The data.gov.uk catalogue still lists only 2023 and 2024; the 2025 files are listed on DVSA’s own open data page.

What these figures cannot tell you

Downloads, citation and press contact

Every table on this page is available as a CSV. Each file opens with five comment lines (starting “#”) giving the source, the licence, the definitions and how to cite; most tools skip them (pandas: comment="#"). They are aggregates only: no registrations, VINs or test ids.

• Models: failure rates by age and mileage (mot-2025-model-table.csv)One row per model: all-ages rate against cars of the same age and mileage, and the first-MOT, age 4 to 5, 7 to 8 and 10 rates, each with its number of tests. 219 rows.
• Models: why they fail, by defect group (mot-2025-model-failure-categories.csv)For each model and each of DVSA’s 13 defect groups: the share of its tests with a defect, the share expected for cars of the same age and mileage, and a flag where the difference is robust. 2,847 rows.
• Models: the specific defects (mot-2025-model-top-defects.csv)For each model, the five most common defects and the five biggest over-indexes against cars of the same age band, in DVSA’s own words. 1,860 rows.
• Postcode areas: results and failure reasons (mot-2025-area-table.csv)One row per postcode area (118 plus XX, merged small areas): initial and outright fail rates, age-adjusted, the share of fails fixed at the station, and every defect group. 119 rows.
• Mileage: failure rate by odometer reading (mot-2025-mileage-curve.csv)For cars aged 6 to 7 and 10 to 11, the initial and outright fail rate in each mileage band, with the tests behind it. 18 rows.
• Mileage: miles driven a year (mot-2025-annual-mileage.csv)Cars aged 5 to 14 by miles driven a year, against the rate expected for their age. 7 rows.

This edition: 2025 MOT tests, definitions version 1.1. This address always holds the 2025 edition; a later year gets an address of its own, so a citation of this page keeps pointing at these figures. The table under “The same tests, both ways of counting” is free to reuse with the credit below.

<iframe src="https://autoza.co.uk/embed/mot-fail-rate" data-autoza-embed title="MOT fail rate by postcode area, 2025 (Autoza)" width="100%" height="820" style="border:0;max-width:640px" loading="lazy"></iframe>
<script src="https://autoza.co.uk/embed.js" async></script>
<p>Source: <a href="https://autoza.co.uk/research/mot-2025">Autoza</a>, from DVSA MOT data</p>

<iframe src="https://autoza.co.uk/embed/mot-fail-rate?chart=model" data-autoza-embed title="First MOT fail rate by model, 2025 (Autoza)" width="100%" height="820" style="border:0;max-width:640px" loading="lazy"></iframe>
<script src="https://autoza.co.uk/embed.js" async></script>
<p>Source: <a href="https://autoza.co.uk/research/mot-2025">Autoza</a>, from DVSA MOT data</p>

<a href="https://autoza.co.uk/research/mot-2025"><img src="https://autoza.co.uk/research/mot-2025/opengraph-image" alt="UK MOT results 2025: which used cars fail, and why. 27.7% of 33,210,823 first MOT tests in Great Britain ended in an initial fail. Source: Autoza analysis of DVSA MOT data, 2025." width="600" height="315" loading="lazy"></a>
<p>Source: <a href="https://autoza.co.uk/research/mot-2025">Autoza</a>, from DVSA MOT data</p>

Read next

• Used cars to avoid in the UK: recalls and MOT failure rates by model
• How each model holds up as the miles mount
• The most reliable used cars, ranked on age-matched MOT results
• What actually fails an MOT
• Checking one car? Our free MOT history check shows its own test record.

Models on this page: fault guides and cars for sale

• Fiat 500 common problems · for sale
• Ford Fiesta common problems · for sale
• Honda Jazz common problems · for sale
• Hyundai i10 common problems · for sale
• Kia Picanto common problems · for sale
• MINI hatch common problems · for sale
• Nissan Micra common problems · for sale
• Peugeot 208 common problems · for sale
• Renault Clio common problems · for sale
• Skoda Fabia common problems · for sale
• Toyota Yaris common problems · for sale
• Vauxhall Corsa common problems · for sale
• Volkswagen Polo common problems · for sale
• Audi A3 common problems · for sale
• BMW 3 Series common problems · for sale
• Ford Focus common problems · for sale
• Ford Kuga common problems · for sale
• Kia Sportage common problems · for sale
• Nissan Juke common problems · for sale
• Nissan Qashqai common problems · for sale
• Peugeot 2008 common problems · for sale
• Vauxhall Astra common problems · for sale
• Vauxhall Mokka common problems · for sale
• Volkswagen Golf common problems · for sale
• Volkswagen Tiguan common problems · for sale
• Nissan Leaf common problems · for sale
• Tesla Model 3 common problems · for sale
• Dacia Duster common problems · for sale
• Mercedes-Benz C-Class common problems · for sale