Every vehicle we check gets a reliability score between 5 and 100. That number is our own work, not something published by the DVSA or DVLA. Because a score is only useful if you understand what went into it, this page documents the entire calculation — every weighting, every penalty, and the places where it can be wrong.
Why publish this? A scoring algorithm you cannot inspect is just a number someone is asking you to trust. We would rather you understood the reasoning and disagreed with it than accepted it blindly.
Why a Score At All?
Raw MOT history is genuinely hard to read. A car with fourteen tests, thirty advisories and two failures might be a well-maintained vehicle with an honest owner, or a tired one that keeps scraping through. Reading that properly means holding several things in your head at once: how often it failed, how serious the failures were, whether things are getting better or worse, and whether the mileage makes sense.
The score is an attempt to do that arithmetic consistently, so that two cars can be compared on the same basis. It is a starting point for a conversation with the seller — not a verdict.
Score Composition at a Glance
The chart below shows the maximum each component can contribute or subtract.
The final result is clamped between 5 and 100, so no single penalty can drive a score to zero.
The Calculation, Step by Step
When full DVSA test history is available, the score is built as follows.
| Step | What happens | Effect |
|---|---|---|
| 1. Pass rate | The proportion of tests passed, multiplied by 0.6. This is the anchor for everything else. | 0 to +60 |
| 2. Consistency bonus | A never-failed record earns a substantial bonus; strong-but-imperfect records earn a smaller one. 100% adds 20, 85% or better adds 10, 75% or better adds 5. | 0 to +20 |
| 3. Recent condition | The last three tests are scored separately, because how a car is doing now matters more than how it did eight years ago. A small trend adjustment nudges the score up if recent results beat the lifetime average, and down if they are worse. | 0 to +15, ±trend |
| 4. Defect severity | Dangerous defects count five times; serious major defects count twice. The total is divided by the number of tests, so a long history is not punished simply for being long. | 0 to −18 |
| 5. Mileage pattern | Readings are normalised to miles and checked in order. Any decrease that is not explained by a unit change triggers a clocking penalty; a clean run earns a small bonus. | −15 or +5 |
| 6. History depth | Eight or more tests adds 5; five or more adds 3. More evidence means more confidence in the result. | 0 to +5 |
| 7. Provenance markers | Where data is available: a stolen marker subtracts 50, a Category A or B write-off 40, a scrapped marker 40, other write-off categories 20, and outstanding finance 15. | 0 to −50 |
Two Decisions That Make the Score Different
Most of the above is arithmetic anyone could reproduce. Two parts involve judgement, and they are the reason our score does not simply mirror the raw pass rate.
Not every major failure is a bad sign
The MOT scheme records a worn tyre and a corroded subframe under the same heading: major. Treating them identically would be misleading. A car that failed on a blown bulb is not comparable to one that failed on structural rot, but the official record gives you no way to tell them apart at a glance.
So we read the defect text and sort major failures into two groups. Items we treat as routine consumables include tyres, bulbs and lamps, wipers, washers, the horn, mirrors, number plates, reflectors and headlamp aim. These still appear in the defect list you see — we never hide them — but they do not drag the score down. Everything else is treated as a serious defect and does affect it.
This is a judgement call, and it is the part of the methodology most open to disagreement. A tyre failure genuinely is a safety issue. Our reasoning is that it says far more about the previous owner's servicing habits than about the fundamental condition of the vehicle, and that a buyer comparing cars cares most about the latter.
A mileage drop is not always clocking
The obvious way to detect odometer tampering is to look for a reading that goes backwards. Done naively, this produces a lot of false accusations, because some vehicles record in kilometres for part of their life and miles for the rest. Switching from kilometres to miles makes the number fall by roughly 38% with nothing untoward having happened.
Before flagging a decrease we therefore check whether the ratio between the two readings sits close to the kilometre-to-mile conversion factor of about 1.609. If it does, we treat it as a unit change rather than tampering and carry on. Where the API does not state the unit at all, we infer it from the shape of the sequence.
The trade-off is deliberate: a genuine rollback that happens to land near that ratio would be missed. We would rather occasionally miss one than routinely accuse honest sellers.
When There Is No MOT History
Some vehicles have no DVSA test history — typically because they are under three years old and have never needed an MOT, or because only DVLA data is available. Scoring these the same way would be meaningless, so a separate and much simpler calculation applies.
It starts from a neutral baseline of 50. A valid MOT status adds 25 and an invalid one subtracts 20. Being taxed adds 15, and not being taxed subtracts 10. Vehicles up to three years old gain 10, and those up to six years old gain 5.
This is a much weaker signal and should be read as little more than a status summary. Where you see a score derived this way, the underlying data simply does not support a reliability judgement.
What the Score Cannot Tell You
A number this compact necessarily throws information away. These are the limits we know about.
- It only sees MOT days. A car can break down repeatedly and still pass every test. Nothing about servicing, clutches, gearboxes, air conditioning or electrical gremlins appears in MOT data at all.
- It cannot see repairs. A failure followed by a proper repair and a failure followed by a bodge look identical in the record.
- It does not know the model. The score compares a car against an absolute scale, not against others of the same make and model. Some models fail more often for reasons of design.
- Low mileage is not scored as good. Very low mileage can indicate long periods standing, which brings its own problems.
- Recent cars score on thin evidence. A car with one test has no meaningful track record, and the history-depth bonus reflects that only weakly.
- Provenance data is not always available. Where it is missing, those penalties simply do not apply — a clean score is not proof of a clean title.
Treat a high score as a reason to keep looking at a car, and a low score as a reason to ask harder questions. Neither replaces a test drive, an inspection, or a paid history check.
Data Sources
Test results, defect text and odometer readings come from the DVSA MOT History API. Vehicle details, MOT status and tax status come from the DVLA Vehicle Enquiry Service. Both are official UK government sources. The scoring, defect classification, unit-switch detection, comparison logic and written analysis on this site are our own.
We do not alter the underlying government data, and where our interpretation differs from the raw record you can always see the raw record alongside it.
Corrections
If you believe the score is wrong for a specific vehicle, or you disagree with how we classify a particular defect, we want to hear about it — that feedback is how the methodology improves. Please get in touch with the registration and what you think is wrong. If the underlying government data itself is incorrect, that must be corrected at source with the DVSA before it will change here.
See it applied to a real vehicle
Enter any UK registration to see the score alongside the full test history it was calculated from.
Check a vehicle