Britain’s MOT records reveal how electric cars age—up to a point

Whether an electric car makes up for the emissions involved in building it depends partly on a number nobody can observe at purchase: how far that car will travel before it leaves the road.

New electric cars create a particularly awkward measurement problem. They have not existed long enough to complete their lifetimes. Researchers can inspect what older cars did, but any conclusion about today’s models must cross a bridge from recorded history to prediction.

A peer-reviewed Nature Communications study published September 21 crossed that bridge using Britain’s unusually extensive vehicle records. Its central finding has two deliberately different halves: first-generation battery-electric vehicles were driven less and did not last as long as conventional cars; if recent improvements continue, battery cars purchased in 2026 are projected to match conventional cars in annual mileage and lifetime.

The first half describes observed history. The second is conditional. “If recent trends continue” is doing real work.

An annual inspection becomes a long-running experiment

Britain’s MOT roadworthiness system leaves behind more than pass-or-fail marks. The official anonymised dataset includes vehicle make and model, test outcomes and odometer readings dating back to the computerisation of the system in 2005.

Link those records across successive years and a car acquires a rough biography: it appeared for inspection, its odometer advanced, and eventually its appearances stopped. Across a fleet, those biographies reveal how much different groups of cars were driven and how long they remained visible on British roads.

That is far better than assuming every vehicle covers an identical lifetime distance. It is not the same as watching each car from factory to scrapyard, however. A missing future inspection does not itself explain what happened. A car might have been scrapped or exported, and newer vehicles have incomplete histories simply because they are still young.

A separate 2025 Nature Energy analysis shows how researchers have handled that uncertainty with survival statistics. It examined more than 264 million MOT results, producing a population of 29.8 million vehicles, but only about 41,600 were battery-electric. Because an exact retirement date was unavailable, that study tested several waiting periods before treating a missed inspection as evidence that a car had left British roads.

Despite using a different analysis, it found the same broad generational pattern: early battery cars had lower reliability, while newer production years improved rapidly enough to approach conventional-car lifespans. That agreement strengthens the case that “electric cars” should not be treated as one timeless technology. Early models and recent ones belong to meaningfully different cohorts.

Administrative data still needs housekeeping. The DVSA warns that a 2018 change to the MOT regime affected the information presented in its files. It has also revised past extracts to restore missing 2017 tests and correct some 2022 fuel-type codes. A huge dataset can reduce sampling noise; it cannot make definitions, revisions or missing observations irrelevant.

The emissions numbers are model outputs, not exhaust readings

The 2026 study reports projected attributional lifecycle emissions of 88 grams of carbon-dioxide equivalent per kilometre for current battery-electric cars, compared with 190 for hybrids and 254–270 for conventional vehicles.

Those figures do not come from an instrument held behind a tailpipe. Battery cars have no tailpipe emissions. Lifecycle accounting instead assigns estimated emissions from stages such as manufacturing and energy use across the distance a vehicle is expected to cover. Mileage and longevity therefore affect the result: a manufacturing burden spread across more kilometres contributes less to each kilometre.

The researchers trained a transformer neural network on the vehicle data to project the mileage and survival of newer cars whose complete lives cannot yet be observed. That makes the reported 88 grams a modeled estimate shaped by assumptions and incomplete histories—not a direct measurement of every current electric car.

A deliberately severe stress check

We can test the published comparison with a transparent calculation. This is not a replacement lifecycle analysis; it is an intentionally pessimistic upper bound.

Suppose both annual mileage and longevity were 10% lower than expected. Lifetime distance would then be:

0.90 × 0.90 = 0.81, or 81% of the baseline.

Now make the harsh assumption that the entire reported 88 g/km behaves like a fixed manufacturing burden. The adjusted figure is:

88 ÷ 0.81 = 108.6 g CO₂e/km

With both mileage and longevity 20% lower, the same calculation gives:

88 ÷ (0.80 × 0.80) = 137.5 g CO₂e/km

Both results remain below the paper’s reported 190 g/km for hybrids and 254–270 g/km for conventional cars. Readers can reproduce the check with any pair of percentages by calculating:

adjusted estimate = 88 ÷ (mileage factor × longevity factor)

This test is stacked against the battery car because electricity-related operational emissions do not really behave like a fixed burden diluted by distance. It also changes only the battery-car estimate while leaving every comparator untouched. A complete sensitivity analysis would vary manufacturing, electricity, fuel, mileage and survival assumptions for every powertrain together. The arithmetic therefore checks whether the published ordering survives one simple stress case; it does not validate the model or prove what any individual car will achieve.

The useful lesson in the MOT records is not that prediction has disappeared. It is that prediction now begins with millions of real odometer histories rather than one assumed lifetime distance. For early electric cars, the records show what happened. For cars purchased in 2026, they show the trend a forecast must extend—and exactly where observation ends.