Explore matched mileage, national transitions and audited roadworthiness in separate workspaces.
Mileage evidence
December 2024–December 2025 · Great Britain · 1,268,479 matched vehicles
Matched mileage cohort · December 2024–December 2025 · spread
Annualised mileage between tests
Miles per year
December 2024–December 2025 · Great Britain · Vehicles matched across the December snapshots
Denominator: 1,268,479 matched vehicles. Observation: Matched vehicle with recorded mileage between normal tests.
Annualised miles between tests · one shared mileage axis
P10
Median
P90
07,50015,000 miles/year
P10 1,748 miles/year
Median 5,981 miles/year
P90 13,803 miles/year
The line spans P10–P90; it is neither a confidence interval nor the full observed range. No unrecorded quartiles or distribution bins are inferred.
Published interval-mileage percentiles; these are not lifetime annualised mileages.
Percentile
Annualised miles between tests
Matched vehicles
P10
1,748
1,268,479
Median
5,981
1,268,479
P90
13,803
1,268,479
Retained support: 1,268,479 matched vehicles; this is not the separate national roadworthiness cohort.
The P10–P90 span describes variation in vehicle use, not statistical uncertainty or claim frequency. The release does not supply a full distribution or percentile confidence intervals.
Definition, reference population & exclusions
Numerator / measure
Percentiles of annualised interval mileage, not a count-based event rate.
Adjustment
Distance between recorded odometer readings annualised over the test interval. No exposure or fuel standardisation is applied in this distribution.
Unit of observation
Matched vehicle with recorded mileage between normal tests
P10 means 10% of this distribution lies at or below that mileage; the median divides it in half; P90 marks 90%. The published P90/P10 ratio is 7.9×. The highest-use 20% account for 45.0% of miles in the matched cohort.
December 2024 snapshot: 2,193,827 vehicles; December 2025 snapshot: 2,247,442. These snapshot totals are not the denominator of the percentile result. Detailed attrition categories and approved histogram bins are not supplied in this summary.
Evidence release: lens-2026-10-09-cbec43fe76c4. Definitions and source limitations remain available in the Mileage workspace.
Definitions and coverage
P10, median and P90 are points in the annualised mileage distribution. The matched cohort compares recorded odometer readings between normal tests. Coverage, fuel and vehicle class definitions differ from the separate roadworthiness build; neither result measures insurance claims.
National mileage transitions
National Class 4 transitions · 2025 · All fuels · All eligible ages · main
National Class 4 records · 2022-01-01–2025-12-31 · Published roadworthiness cohort
Two consecutive observed mileage intervals, reconstructed from three normal tests. Rows describe earlier use; columns describe later use. Both intervals use the same vehicle.
Movement between annual mileage bands
Share of each earlier-use band (%)
Final test year 2025 · Great Britain · National Class 4 cohort · all supported fuels · all eligible ages
Denominator: Each earlier-mileage row’s observed transitions; 24,881,848 across all rows. Observation: One observed pair of successive mileage intervals from three normal tests.
0%25%50%75%100%Hatched: no observations or withheld
Rows: earlier annualised miles. Columns: later annualised miles. Percentages use the row count; 0.0% is an observed zero. Unsupported rows have no published percentage.
Earlier ↓ / Later → miles/year
0–<5,000
5,000–<10,000
10,000–<15,000
15,000–<20,000
20,000+
Row count
0–<5,000
80.0%8,729,760 transitions
16.1%1,753,523 transitions
2.7%298,041 transitions
0.7%79,749 transitions
0.4%48,687 transitions
10,909,760Supported
5,000–<10,000
22.3%2,052,336 transitions
64.3%5,909,746 transitions
11.2%1,030,889 transitions
1.6%143,799 transitions
0.6%58,214 transitions
9,194,984Supported
10,000–<15,000
8.0%267,874 transitions
34.6%1,160,122 transitions
47.0%1,572,827 transitions
8.5%286,359 transitions
1.9%62,227 transitions
3,349,409Supported
15,000–<20,000
6.6%61,863 transitions
16.2%152,855 transitions
33.7%316,872 transitions
33.8%318,378 transitions
9.7%91,047 transitions
941,015Supported
20,000+
6.9%33,634 transitions
11.0%53,574 transitions
14.5%70,395 transitions
21.3%103,885 transitions
46.3%225,192 transitions
486,680Supported
Rows: earlier annualised miles. Columns: later annualised miles. Percentages use the row count; 0.0% is an observed zero. Unsupported rows have no published percentage.
Earlier ↓ / Later → miles/year
0–<5,000
5,000–<10,000
10,000–<15,000
15,000–<20,000
20,000+
Row count
0–<5,000
80.0%8,729,760 transitions
16.1%1,753,523 transitions
2.7%298,041 transitions
0.7%79,749 transitions
0.4%48,687 transitions
10,909,760Supported
5,000–<10,000
22.3%2,052,336 transitions
64.3%5,909,746 transitions
11.2%1,030,889 transitions
1.6%143,799 transitions
0.6%58,214 transitions
9,194,984Supported
10,000–<15,000
8.0%267,874 transitions
34.6%1,160,122 transitions
47.0%1,572,827 transitions
8.5%286,359 transitions
1.9%62,227 transitions
3,349,409Supported
15,000–<20,000
6.6%61,863 transitions
16.2%152,855 transitions
33.7%316,872 transitions
33.8%318,378 transitions
9.7%91,047 transitions
941,015Supported
20,000+
6.9%33,634 transitions
11.0%53,574 transitions
14.5%70,395 transitions
21.3%103,885 transitions
46.3%225,192 transitions
486,680Supported
Retained support: 5 of 5 rows meet the minimum of 100 transitions; 24,881,848 / 24,881,848 transitions are in supported rows.
Conditional on observing three usable tests. Missing tests are not zero mileage or scrappage. This selected cohort does not estimate movement for all national vehicles, and no transition confidence intervals are estimated here.
Definition, reference population & exclusions
Numerator / measure
Transitions from the earlier row band into the later column band.
Adjustment
No standardisation. Both observed intervals belong to the same vehicle. Eligibility: 180–730 days per interval.
Unit of observation
One observed pair of successive mileage intervals from three normal tests
67.3% remain in the same mileage band (16,755,903 / 24,881,848 observed transitions across all rows). Each supported row sums to 100% before rounding. The colour scale is fixed at 0–100% across selections.
Three consecutive observed completed normal Class 4 tests; two non-decreasing odometer intervals of 180–730 days (270–450 sensitivity); unique test dates at all endpoints; same eligible fuel; one existing cohort anchor per vehicle/year. Five-thousand-mile annualised bands with a 20,000+ tail. Non-return and missing observations are unresolved.
Build coverage: 2022-01-01–2025-12-31. Early years have less prior test history. The full-build eligibility register below reports exclusions before these display filters.
Both rows keep final test year 2025, all supported fuels, the selected age group and the same mileage bands. The narrower window selects a subset of observed transitions; its denominators and vehicle mix can change.
Existing interval sensitivity, without re-estimation. Same-band share uses all transitions in that window and is withheld below 100; individual matrix rows require 100 each.
Both intervals must span
Observed transitions
Share of main cohort
Remain in same band
Supported matrix rows
Inspect
180–730 days · main
24,881,848
100.0%
67.3%
5 / 5
270–450 days · sensitivity
22,605,260
90.9%
68.7%
5 / 5
These overlapping samples are not independent replications. Differences are descriptive and may reflect eligibility and vehicle mix; no confidence interval or significance test is supplied.
Full-build eligibility and review
Three consecutive observed completed normal Class 4 tests; two non-decreasing odometer intervals of 180–730 days (270–450 sensitivity); unique test dates at all endpoints; same eligible fuel; one existing cohort anchor per vehicle/year. Five-thousand-mile annualised bands with a 20,000+ tail. Non-return and missing observations are unresolved.
fewer than three observed tests: 63,188,874 cohort anchors
These controls cover the full build, before the display filters above. Review: Johan Strydom, 2026-09-26.
Conditional on three observed usable tests, not the national vehicle stock or every insured driver.
Vehicle identity follows the DVSA source; no cross-source or household linkage.
Age is measured at the final anchor. Annualised use is an interval estimate, not a full calendar-year mileage total.
Overlapping histories share vehicles; do not treat transitions as independent insurance claims or infer scrappage.
Both intervals share the middle odometer reading; measurement error can create apparent reversals in mileage.
Non-decreasing odometer jumps remain in the main estimate; the 20,000+ band is an open tail.
Early source years have limited prior histories and can overrepresent frequent testers; changes in coverage are not a population trend.
Audited roadworthiness
Comparison · 2025 · Battery electric / Petrol · initial failures · Age & mileage adjusted
What changed since the previous approved release?
Compare this exact selection. Live source checks are shown separately from approved data, methods and results.
Reviewed analytical build
2022-01-01 to 2025-12-31 · 35,879,674 unique vehicles · 117,537,248 vehicle-years
Release and build identifiersEvidence release: lens-2026-10-09-cbec43fe76c4 · Build: 719eec692a459667
Presets
Presets set age/mileage adjustment and keep your selected year and failure definition.
Shared by A and B:2025 · Great Britain · Class 4Initial (F + PRS) failures per vehicle-yearAge & mileage adjusted
Initial failure: raw and adjusted
Failures per 100 vehicle-years (%)
2025 · Great Britain · Normal Class 4 MOT tests; ages 3–30
Denominator: Battery electric: 381,053 observed vehicle-years · Petrol: 16,049,996 observed vehicle-years Observation: One first completed normal test per vehicle per year.
Difference basis: Age & mileage adjusted. Raw points include all observed vehicle-years. Filled points use age & mileage adjusted common support; changing basis can change that population.
Hollow points: raw rates with 95% Wilson intervals. Filled points: adjusted rates with 97.5% conservative bounds. Bounds use two decimal places.
A · Battery electric381,053 observed vehicle-years
Shared 0–100% axes. Exact displayed values and evidence states are also available in Table.
Age & mileage adjusted · A (Battery electric) minus B (Petrol): 2.2 percentage points · conservative interval 1.5 to 3.0 pp (at least 95% coverage)
−5 pp0+5 pp
Percentage-point difference, not relative percentage change.
Raw rates use all observed vehicle-years; adjusted rates use common support and fixed reference weights. Both rows retain the same year and failure definition.
Cohort
Basis
Rate
Rate interval
Retained / all vehicle-years
Retained support
State
A · Battery electric
Raw observed
15.9%
15.74%–15.98% (95% Wilson)
381,053 / 381,053
100.0%
Available
A · Battery electric
Age & mileage adjusted
24.0%
23.33%–24.77% (97.5% conservative)
370,464 / 381,053
97.2%
Available
B · Petrol
Raw observed
28.6%
28.54%–28.58% (95% Wilson)
16,049,996 / 16,049,996
100.0%
Available
B · Petrol
Age & mileage adjusted
21.8%
21.76%–21.85% (97.5% conservative)
10,548,831 / 16,049,996
65.7%
Available
Age & mileage adjusted · A (Battery electric) minus B (Petrol): 2.2 percentage points · conservative interval 1.5 to 3.0 pp (at least 95% coverage)
−5 pp0+5 pp
Percentage-point difference, not relative percentage change.
Raw rate intervals use Wilson 95%; adjusted fuel bounds are conservative 97.5%. Combining adjusted endpoints gives at least 95% coverage for the pairwise difference under within-year vehicle independence. These intervals do not remove selection bias, vehicle-mix confounding or establish insurance claim risk.
Definition, reference population & exclusions
Numerator / measure
Initial failures (F + PRS). Raw rates divide failures by all selected vehicle-years. Adjusted rates weight within-stratum proportions; they are not pooled failures / vehicles ratios.
Adjustment
Selected difference: Age & mileage adjusted. The chart pairs raw rates with age & mileage adjusted rates. Adjusted rates use fixed 2025 pooled reference weights, with at least 100 vehicles per fuel in every released year (2022, 2023, 2024, 2025).
Unit of observation
One first completed normal test per vehicle per year
Initial failures include passes after rectification (PRS). Retests are excluded. Vehicles with inconsistent fuel codes are excluded before the displayed cohort is formed.
Adjusted reference: 10,919,295 of 16,431,049 pooled vehicle-years in 2025 (66.5%). This differs from each displayed year’s fuel-specific support. Selecting Raw changes the difference to an unadjusted all-observation comparison; the adjusted chart context remains labelled.
Support exclusions within the reviewed selected cohort. Earlier source-quality exclusions are recorded under Data quality.
Fuel
All vehicle-years
Outside common support
Retained vehicle-years
Observed failures in retained rows
Battery electric
381,053
10,589
370,464
57,343
Petrol
16,049,996
5,501,165
10,548,831
2,329,702
Inspect 40 fixed reference weights
2025 pooled reference. Weights sum to 100% before rounding.
Both cohorts keep 2025, Great Britain, the initial failure definition and vehicle-year units. Within each adjusted comparison, A and B use the same fixed 2025 reference and common strata across every released year.
Each row is a separate governed comparison. Moving between rows changes support and weights; the change is not solely the effect of adjustment.
10,919,295 reference vehicle-years · 40 common strata
Raw rates use all selected vehicle-years. Age and age/mileage restrictions can retain different vehicles. Changing a fuel also recalculates the pair’s common support; changing only the year keeps each adjusted comparison’s reference weights fixed.
Definitions, uncertainty and limits
Coverage starts no earlier than 2022, after DVSA mileage-unit and fuel classification changes. Class 4 is broader than passenger cars. Each vehicle contributes one first completed normal test per year; partial and full retests are excluded from the main rate. This differs from all-test statistics and must not be presented as a direct reproduction without cohort alignment.
Raw and individual-cell intervals use Wilson 95% intervals. Adjusted rates use one frozen reference population and strata with at least 100 vehicles in both fuels in every year. Conservative distribution-free intervals allow a pairwise difference with at least 95% coverage under within-year vehicle independence. Precision does not remove selection bias, make/model confounding or source errors. All intervals are pointwise; exploratory comparisons are not multiplicity-adjusted.
MOT roadworthiness does not measure battery state of health, breakdowns, repairs between tests, software reliability, warranty cost or insurance claim risk. This Lab tests hypotheses; it cannot guarantee a preferred conclusion.
Distance travelled is one dimension of exposure. MOT mileage does not identify where or when a vehicle was driven, the keeper’s address, battery health or insurance losses.