That used EV says its battery is healthy. The number may not mean what you think

A used electric car can present a reassuring battery state-of-health figure and still leave the buyer with the wrong impression. New research comparing vehicle-reported health against charging data found real capacity differences of 12% to 25% within every model group studied, while the onboard figures frequently failed to reflect those gaps.

That finding arrives while the used market is still asking the same nervous question. In an August survey by Startline Motor Finance, 57% of potential buyers said they would not buy a used EV without an independent battery check. They are asking the right thing. The harder problem is which number deserves the trust they want to put in it.

The March 2026 preprint followed 1,114 EVs in South Korea for 375 days. The fleet covered Hyundai, Kia, Genesis, Audi and Volkswagen vehicles, although those five brands belong to only two corporate groups. Researchers estimated relative capacity independently by measuring how much charge entered a pack through controlled, constant-current portions of charging sessions, keeping current variation below 25% and battery temperature between 10°C and 35°C.

They then compared that evidence with state of health, or SOH, reported by the battery-management system. Results varied sharply by platform. In the commercial Porter and Bongo EV group, a measured capacity gap of 24.7% separated usage quartiles, yet average reported SOH remained essentially 100% in both. A 192-cell E-GMP group showed a 16.4% measured gap, but only a one-percentage-point SOH difference. For the Niro and Kona group, a 20.7% capacity gap corresponded to 1.6 points of reported SOH.

The remaining 384 vehicles did not expose SOH at all, including every Audi and Volkswagen MEB model in the sample. Those packs were invisible to the diagnostic interface the researchers used.

This is a preprint, not a peer-reviewed verdict on every EV. Its cars were self-selected, all operated in South Korea, and its manufacturer coverage was narrower than the five badges suggest. The authors also used a partial-charge capacity proxy rather than removing and laboratory-testing every pack. They did, however, validate the method against wider charging windows and an identical 55.6 Ah pouch cell in controlled testing.

State of health sounds standardized because it is usually displayed as a percentage. It is not necessarily standardized in meaning. Each manufacturer can use a proprietary algorithm, and the original reference may involve gross pack capacity, usable capacity, calibration choices and hidden buffer. Ninety-five percent from one model is not automatically comparable with 95% from another.

Efficient Alex makes the same distinction in a battery-report walkthrough: work from usable capacity, not the larger marketing figure. A test can be internally consistent and still produce a misleading degradation percentage if the wrong new-car baseline is used.

An owner with an OBD-II dongle may retrieve cell voltages, temperatures and a BMS health value. UpsideDownFork shows that inexpensive route on a used MG5, but the new study explains its limit: reading a value accurately is not the same as proving that the value tracks remaining energy accurately.

A fuller measurement counts energy delivered over a meaningful state-of-charge window. On the InsideEVs Plugged-In podcast, Recurrent's Scott Case described Tesla's home self-test in those terms: wait until the car is below 20% charge, plug it in, let it dump the rest as heat, then refill to 100% and read how many kilowatt-hours went in. He ran it on a 2023 Model Y and got 92% or 93% — then noted that the same procedure is harder to obtain outside Tesla. That is the access gap the Korean fleet study is measuring from the other side.

A useful battery report should identify the testing method, the usable-capacity baseline, starting and ending charge levels, battery temperature and the date. A bare SOH percentage without those details is closer to a claim than an inspection.

Ask the seller for the manufacturer diagnostic report, but do not stop there. Check cell-voltage imbalance and stored fault codes, confirm that the car can fast-charge normally, and compare energy added during a controlled charge with the change in indicated state of charge. A road test from a known charge level can still work as a sanity check, not as a substitute for a documented capacity measurement. No single shortcut replaces model-specific knowledge, because pack buffers and reporting behaviour differ.

Electrifying's used-battery explainer puts a useful brake on the panic: normal degradation is usually gradual, and fleet evidence does not support treating every older pack as a looming replacement bill. The issue raised by the preprint is not that used EV batteries are broadly failing. It is that healthy and worn examples can be difficult to distinguish when the available gauge is opaque.

Regulation is moving toward disclosure. The EU battery regulation's 2027 SOH transparency rules, and California's Advanced Clean Cars II durability guarantees, both assume the reported percentage is meaningful. Disclosure will not solve the problem if two identical percentages describe different physical conditions. Buyers need a repeatable measurement, its assumptions and enough context to compare like with like. Until that exists at every dealer, the safest reading of a used EV's SOH figure is evidence to investigate, not a certificate to accept.