VehicleScore research
Mileage discrepancies in MOT history
Of 1,136,102 vehicles with at least two valid mileage readings, 55,875 had a later reading that was lower than an earlier one. That is 4.9%. A lower reading is a discrepancy in the record. It is not, on its own, evidence that someone altered the mileage.
How large was the drop?
47,574 vehicles were lower by more than 100 miles (4.2% of vehicles with two readings). 41,695 were lower by more than 1,000 (3.7%). 34,957 by more than 5,000 (3.1%). 30,285 by more than 10,000 (2.7%).
Among vehicles with any drop, 54.2% had their largest drop over 10,000 miles. That is 30,285 vehicles. A five-figure drop looks dramatic. The next section is why VehicleScore does not treat it as proof of tampering.
Largest drop, for vehicles that had at least one later reading lower than an earlier one. Shares are of those vehicles, not of every car with two readings.
| Largest drop | Vehicles | Share of drops |
|---|---|---|
| 1–100 miles | 8,301 | 14.9% |
| 101–1,000 miles | 5,879 | 10.5% |
| 1,001–5,000 miles | 6,738 | 12.1% |
| 5,001–10,000 miles | 4,672 | 8.4% |
| Over 10,000 miles | 30,285 | 54.2% |
Some of the big drops are followed by the mileage returning
Of the 41,695 vehicles with a drop over 1,000 miles, 12,511 later had a reading back to at least 98% of the figure from before the drop. That is 30.0%. The rule requires this of every drop over 1,000 miles on that vehicle, not just one of them.
Another 4,249 vehicles, 10.2% of the over-1,000 group, had a drop that landed within 20% of one tenth of the previous reading. A mileage that falls from about 80,000 to about 8,000 has the shape of a digit missed at the keyboard. It also has other possible explanations. The label is a pattern. It is not a finding that anyone altered an odometer.
VehicleScore has not published a model ranking of mileage discrepancies. The model table in the census is not adjusted for age. Older vehicles have more readings and a much higher discrepancy rate, so an unadjusted “most clocked cars” list would mostly rank older models. Publishing it would describe age, then dress the result up as tampering. We are not doing that.
The discrepancy that matters is on one registration
A dataset rate cannot tell you whether the car you are buying has a mileage gap. Enter the registration. The free check shows the MOT mileage history. A paid Hidden History check is a separate product and is not what displays the MOT mileage list.
Age dominates the rate
The age here is the vehicle’s age at its latest mileage reading, taken from the first-used date. Each vehicle is in one band. This is not the overlapping age-at-test count used on the MOT pass rate by age page.
A drop of more than 1,000 miles was recorded for 0.6% of vehicles aged 3–5 at the latest reading (626 of 111,404). At age 15 or over the rate was 7.8% (23,714 of 303,193). The young-car rate is not a clean zero: ages 0–2 are 0.8% on only 4,996 vehicles with two readings.
Vehicles with a mileage drop over 1,000 miles, by age
Denominator: vehicles with two valid readings and a usable first-used date. 1,131,318 of 1,136,102 vehicles with two readings fell into a band.
| Age at latest reading | Vehicles with two readings | Drop over 1,000 miles | Rate |
|---|---|---|---|
| 0–2 years | 4,996 | 41 | 0.8% |
| 3–5 years | 111,404 | 626 | 0.6% |
| 6–8 years | 211,034 | 2,488 | 1.2% |
| 9–11 years | 276,810 | 5,907 | 2.1% |
| 12–14 years | 223,881 | 8,042 | 3.6% |
| 15 or over years | 303,193 | 23,714 | 7.8% |
More readings create more chances for a later figure to be lower, including chances created by a typing error that a subsequent test corrects. That is a reason the age gradient exists. It is not a complete account of it. The file cannot see an odometer.
The mileage check is the right place to read one car’s sequence. This page is the reason a single backwards reading should be read next to the readings around it.
Related VehicleScore research
Methodology
Snapshot 27 September 2026, job 2026-09-27.1. A valid reading is a positive mileage up to 1,000,000. A discrepancy is any later reading lower than an earlier one on the same vehicle. Thresholds are cumulative: “over 1,000” includes the vehicles that were also over 10,000. Drop-size buckets use the largest drop and do not overlap.
Isolated recovery means every drop over 1,000 miles was followed by a later reading at least 98% of the pre-drop reading. The one-tenth pattern means a drop over 1,000 miles landed within 20% of a tenth of the previous reading. Both are pattern labels.
Age uses the latest reading timestamp minus the first-used date. 4,784 vehicles with two readings had no usable age and appear only in the overall rate. No registration is published. The sample is vehicles checked through VehicleScore.