This page exists so that nothing on this site has to be taken on trust. Every formula is written out, every default is sourced, and every value we model rather than measure is labelled as an estimate wherever it appears.
The three rules everything follows
- Published figures are stored exactly as published. EPA publishes miles, WLTP publishes kilometres, and we keep both in their native units. Converting an EPA figure into kilometres for display is fine — it is the same measured quantity — but we never present a converted EPA number as a WLTP number, and we never average the two.
- Money is never converted. A US price and a European price are two different market prices set by two different companies, not one price at an exchange rate. Where a car is not sold in a market, that market’s price is blank rather than estimated.
- Modelled values are labelled. Anything carrying an est tag is calculated by us from a stated formula, not measured by anyone. The formulas are all on this page.
Where the specification data comes from
Battery capacity, WLTP range, observed real-world range, consumption, charging power and charge times, kerb weight, acceleration and European and UK pricing come from EV Database, which maintains a consistent methodology across manufacturers. US range and consumption figures come from the EPA via fueleconomy.gov. Crash ratings come from Euro NCAP and NHTSA directly. Every individual figure on a vehicle page carries a link to its own source and the month we recorded it.
Where a figure has no citable source, the row is left blank. We would rather show a dash than a plausible-looking number nobody can check.
Range: four different numbers, four different meanings
| Figure | What it is | How to read it |
|---|---|---|
| EPA | US regulatory combined cycle | The most conservative official test. Closest to real driving of the three. |
| WLTP | European regulatory combined cycle | Typically 15–25% higher than EPA for the same car. Useful for comparing European cars to each other, not for planning a journey. |
| CLTC | Chinese regulatory cycle | The most optimistic of the three. Shown for completeness only. |
| Observed real-world | Independent mixed-conditions testing | The number that decides whether a journey needs a charging stop. |
Charging: why we lead with the 10–80% time
Peak DC power is a marketing figure. Many cars hold their advertised peak for well under two minutes and only inside a narrow state-of-charge window, at a battery temperature they may not reach in winter. The time from 10 to 80 percent is what you actually stand around for, and it is the row we highlight. Pack voltage is the useful supporting detail: 800-volt architectures move the same power at half the current, generate less waste heat, and therefore hold high power for far longer before tapering.
Battery capacity loss
This is the field where it would be easiest to invent a number, so it is worth being precise about what we know and what we do not.
What is published: Geotab’s 2026 study of more than 22,700 electric vehicles across 21 models found an average capacity loss of 2.3% per year, projecting to roughly 81.6% of original capacity after eight years. That is up from 1.8% in their 2024 analysis, which Geotab attributes largely to increased reliance on high-power DC fast charging. Vehicles doing more than about 12% of their sessions on DC above 100 kW degraded at up to 3.0% per year, against roughly 1.5% for cars charged mainly on AC.
What is not published: per-model figures. Geotab anonymises the models in its findings, no manufacturer releases fleet degradation data, and no independent body measures it at the necessary scale. Anyone publishing a model-by-model degradation table is estimating.
What we therefore do: we use Geotab’s own class split — 2.0% per year for cars, 2.7% for SUVs and multi-purpose vehicles — and label the source on every vehicle page as a class average rather than a measurement of that car. Two SUVs on this site show the same figure because we genuinely do not know which of them degrades faster. Treat it as an expectation for the class, not a property of the model.
Loss is expressed per year rather than per kilometre because that is how the underlying data is measured, and because calendar age drives a substantial part of it independently of mileage.
Projected state of health
The eight-year projection and the curve on each vehicle page are a straight line: capacity = 100% − (annual loss × years). Real degradation is not linear — it is faster in the first months and then flattens — so this is a deliberately simple, slightly pessimistic model rather than a curve fitted to data we do not have. It carries an estimate tag everywhere it appears.
The battery safety score
This is the only number on the site that is ours rather than a source’s. It is computed, not assigned, so the same inputs always produce the same score and nobody can quietly nudge a car up or down. Four components, worth 2.5 points each:
| Component | Scoring | Why it is in here |
|---|---|---|
| Thermal management | Direct refrigerant or active liquid 2.5 · forced air 1.0 · passive air 0 | The strongest single predictor of whether a pack stays inside its safe temperature window under fast charging and hard driving. |
| Cell chemistry | LFP and solid-state 2.5 · LMFP 2.3 · sodium-ion 2.2 · NMC and NCMA 1.5 · NCA 1.2 | Iron-phosphate cells begin thermal runaway at a far higher temperature than nickel-rich cells, and release less energy when they do. |
| Crash rating | Best of NHTSA or Euro NCAP overall, scaled: (stars ÷ 5) × 2.5 | Independent structural testing. The better of the two is used because most cars are tested by only one body. |
| Recall record | Starts at 2.5 · −0.6 per battery or high-voltage recall · −1.5 more for a documented thermal runaway campaign · floor of 0 | Full marks until something has actually gone wrong in the field. |
The score is the points earned divided by the points available across whichever components we have data for, scaled to 10. If fewer than three of the four are available, no score is shown at all — an incomplete score is worse than none. Every score displays its coverage, so a 9.2 from three components is visibly not the same claim as a 9.2 from four.
The obvious limitation: a car with an excellent chemistry and cooling system but no recall history recorded here can score very highly on three components. Read the coverage figure alongside the score.
Estimated battery replacement cost
Formula: usable capacity (kWh) × $165, rounded to the nearest hundred. The rate is the midpoint of the roughly $130–$200 per kWh that out-of-warranty pack replacements were quoted at in 2026, a figure that includes both the pack and the labour. The euro estimate applies the same per-kWh rate at approximate parity.
Three honest caveats. Dealer quotes vary enormously for the same car. Remanufactured and third-party packs undercut official prices by thirty to fifty percent where they are available. And full pack replacement is rare in practice — module-level repair is increasingly the norm and costs a fraction of these figures. Treat this row as an outer bound on a risk, not a bill you should expect.
Running cost calculator: every default and its source
All of these are starting values, visible and editable in the calculator itself. If one is wrong for you, change it and the answer moves.
| Input | US | EU | UK | Source |
|---|---|---|---|---|
| Home electricity | $0.184/kWh | €0.2629/kWh | £0.26/kWh | EIA (Sept 2026); Eurostat household price, H2 2025, 5,000–15,000 kWh/year band — the band a household that charges a car sits in (the headline band is €0.2896); Ofgem price cap, approximate |
| Public rapid charging | $0.48/kWh | €0.59/kWh | £0.79/kWh | Typical network rates. These vary more than any other input on this list. |
| Petrol | $4.07/gal | €2.09/L | £1.35/L | EIA weekly retail, 31 Aug 2026; EU Weekly Oil Bulletin, consumption-weighted EU27 average for Euro-95, 21 Sept 2026 (the unweighted average of the 27 national prices is €1.96; national prices range from about €1.34 to €2.62) |
| Grid carbon | 350 g/kWh | 230 g/kWh | 131 g/kWh | EPA eGRID 2023 national average; Ember yearly electricity data (approximate); DEFRA/DESNZ 2026 conversion factors |
| Petrol carbon | 8.887 kg/gal | 2.31 kg/L | 2.31 kg/L | EPA greenhouse gas equivalencies; standard combustion factor |
| Annual distance | 13,500 mi | 13,000 km | 11,000 km | National averages, rounded |
What the calculator counts
Purchase price after any incentive you enter, electricity at a blend of your home and public rates weighted by the split you set, petrol at the consumption you enter, and depreciation using the estimated five-year residual where we have one. Energy consumption uses the market-appropriate published figure: EPA kWh per 100 miles in imperial mode, and EV Database’s real-world Wh per kilometre in metric — not the WLTP laboratory value, which is lower for every car.
What it does not count, and why
Insurance, tyres, servicing, road tax and finance interest are excluded. They differ enough between drivers, countries and insurers that a national average would make the answer look more precise than it is — and a false precision is worse than an acknowledged gap. Servicing does generally favour the electric car; tyres generally do not, because the extra weight wears them faster.
The carbon figure is generation and tailpipe only. It excludes battery manufacturing, where an electric car starts at a deficit and catches up over the first years of driving — sooner on a clean grid, later on a coal-heavy one. Anyone quoting an electric car as zero-emission from day one is not counting properly, and neither is anyone claiming the manufacturing debt is never repaid.
Independence
We do not sell cars, we do not broker finance, and no manufacturer pays for placement, ranking or a rewritten verdict. Where a car does badly on a metric, the page says so.
Found something wrong? The corrections policy explains how we handle it, and the contact page is the fastest way to tell us.
Who writes this, and what happens when it is wrong
The assumptions above are only worth as much as the people applying them, so both are published: who runs this site and how it is funded, and the editorial policy that says what we will and will not do for money — no manufacturer funding, no paid placement, and no ranking that can be bought. If you find a figure that is wrong, the corrections policy sets out how it gets fixed and what we publish about the fix, and contact is the fastest way to tell us.