How to measure your Tesla's battery degradation
There are three honest ways to measure it, and each one moves from day to day by more than the pack wears in months. Reading through that is most of the skill.
12 November 2026 · 8 min read
Degradation is the capacity a battery has lost since it was new. No gauge on a Tesla shows it. You can estimate it three ways, and each estimate moves around more from one day to the next than the pack itself does in months. Knowing why is what separates a measurement from a scare.
Method 1: Tesla's own battery health test
On supported models and software versions, the car can test itself, under Controls → Service → Battery Health. Plugged in to an AC charger, it runs the pack down to nearly empty and charges it back to 100%, then reports the energy the battery holds as a percentage of what it held when new. It asks for the car to be parked, at or below 20% charge, on an AC charger that delivers at least about 5 kW, with no update pending. Expect it to take most of a day, and to be allowed about once every six months.
It is the closest thing to a measurement an owner can get, because it does what the other methods can only infer: it puts a full cycle through the pack and counts. Its limits are practical. It is rare, it leaves the car sitting at 100% afterwards, and it is not offered on every model and software version. And it produces one number per test — where the pack is, not which way it is going.
Method 2: rated range at 100%
The range a car shows at full charge falls as the pack ages, so the classic method is to charge to 100% now and then and write the figure down. You do not need to charge to 100%: divide the range shown by the charge level and you have a reading on any day. 300 km at 80% extrapolates to 375 km at 100%; the same works in miles.
The catch is that the range figure is not a measurement of the pack. It is the car's estimate, built on what the battery management system (BMS) currently believes the pack holds. When that belief drifts, the range drifts with it, and when the BMS recalibrates, the range can jump back. That is why range lost over a winter sometimes returns in spring, and why the health test can shift it.
One reassurance from our own data: the charge level you read it at barely matters on average. Across three cars, the median extrapolation from readings below 30% charge landed within 1% of the one from readings above 80%.
Method 3: the energy a charge puts in
The third way skips the range estimate. Take a charge where you know both the energy added and the battery level at each end, and divide: 45 kWh that took the battery from 20% to 80% implies a 75 kWh pack. Repeated over many charges, that gives usable capacity in kWh, which you can set against the published figure for your car.
Single sessions are noisy here too. The energy counter is the car's own, the charge level is a whole percent, and a session that starts while the counter still carries the previous one's figure can imply a pack half as big again. Across 244 charging sessions on four cars logged by evlogr, each adding at least 15 percentage points, the gap between the 10th- and 90th-percentile session was between 6% and 23% of that car's median. Any one session can be wrong by more than the degradation you are trying to find.
We learned that the hard way. We estimate each car's usable capacity like this, and the estimate used to come from the single largest charge in the newest batch of data — so it moved by more in a week than a pack wears in a year. It now takes the median of every qualifying charge the car has given us, needs at least five before it will overrule the published figure for the model, and is never allowed above that figure: an estimate above spec is charging loss leaking into the arithmetic, not a battery that grew.
Why a single reading swings
- Calibration. The BMS estimates capacity from voltage and current, and the estimate wanders between recalibrations. Weeks spent between 40% and 80%, never near full or empty, give it less to anchor on.
- Temperature. A cold pack holds energy back, and both the range estimate and the charging figures move with it. Winter readings can look like degradation that spring gives back.
- Rounding. The charge level is a whole percent. At 20% charge, half a percent of rounding is 2.5% of the extrapolated figure.
- One-off sessions. A counter carried over from the last session, a charge interrupted halfway, a short top-up — each produces a number the next one contradicts.
How large is that in practice? We measured it on the three cars evlogr has logged for two months or more:
| What moves | By how much |
|---|---|
| Same car, same day: highest vs lowest reading | 1.6% typically; 3.3% or more on 1 day in 10 |
| Daily figure, one day to the next | 0.26% typically; 1.15% or more on 1 day in 10 |
| Daily figure, highest vs lowest day over the period | 2–7%, depending on the car |
Pick a good day and a bad day from that period and you could “measure” several percent of degradation that is mostly noise. A chart is only useful if it lets you see past that.
How to read through the noise
- Watch a trend over months, not a reading. Compare monthly medians, or the shape of a line through many weeks.
- Compare like with like: similar outside temperature, similar charge level, the same season a year apart.
- Take one reading per day, the same way every time, so a day with a dozen low-charge readings does not count a dozen times.
- Treat a sudden jump after a full charge or a health test as recalibration, not damage or repair.
- If the health test is available to you, use one result as an anchor and the logged trend to see which way the pack is moving between tests.
What evlogr's battery page shows
The battery page charts the estimated range at 100% charge over time: the car's reported range divided by its charge level, read every time we sample the car, keeping one figure per day — that day's highest. Beside the chart are the latest figure, the first one we have, and the change between them. That change compares two single days, so it carries the day-to-day swing above; read the line rather than the number.
The usable-capacity estimate from method 3 is shown with your vehicle in Settings, marked as auto-detected when it comes from your charges, or as the model spec when the published figure decided it. Battery-health trends are part of Pro; the charge-level timeline is free. The line starts the day you connect, because imported history carries no range readings.
These figures come from evlogr, which logs a Tesla’s drives and charges automatically and prices each charge against your own time-of-use tariff. Free on one car.