How to Check Battery Health When Buying a Used EV

Ask for ninety minutes instead of thirty, and say what the extra hour is for. The seller will assume you want a longer drive. What you actually want is to check the battery health of a used EV the only way an ordinary buyer can: the car standing still at a fast charger with its charging screen lit, and two VIN lookups running while it charges.

That request changes the appointment more than anything else on this page. A used EV's drivetrain has almost nothing in it to go wrong in the way a used engine does, so the ordinary test-drive ritual, the listening for noises and the watching of temperature gauges, is aimed at a car that is not in front of you. What can be wrong is expensive and quiet: a pack that has lost more capacity than the mileage suggests, a charging system that will not take power at a useful rate, an unrepaired battery recall, and a warranty with less time on it than the model year implies.

None of those show up on a drive around the block.

The range figure on the dash is the seller's least useful number

It will be the first thing quoted to you. Shows 246 at full. That estimate is recomputed from recent driving, recent temperatures and recent climate control use, and it will move by twenty miles on a pack that has not changed. On a car that has been sitting on a forecourt being moved fifty yards at a time for three weeks, it is close to meaningless.

The comparison worth making is against the EPA figure for that exact year and trim, and that is published. The Fuel Economy database run by the EPA and the Department of Energy returns, for a 2021 Chevrolet Bolt EV, a combined range of 259 miles, 118 MPGe combined, an energy use of 28.7 kWh per 100 miles and an estimated 10.0 hours to charge on 240 volts (fueleconomy.gov vehicle record 43663, read 8 September 2026). Those figures are the baseline the car was rated against when new.

Do not then subtract. A Bolt showing 230 miles at full charge is not announcing eleven percent degradation, because the two numbers are not measuring the same thing. What the EPA figure gives you is a sanity check with a wide band: a five-year-old car predicting somewhere near its rated range is unremarkable, and one predicting half of it is a question you ask out loud. The actual capacity figure lives elsewhere, and where it hides on each brand, why two apps disagree about it, and what a 2026 rule changed is a longer story than a test drive has room for. Read that before the appointment, not during it.

Arrive with the pack low, or you have wasted the trip

Here is the mistake that quietly ruins the visit, and it is usually made by a helpful seller rather than by you.

A lithium pack takes its highest DC power in the lower part of its state of charge and tapers as it fills. Plug in at 85 percent and the car will draw a modest fraction of what the stall can deliver, and it will do that whether the pack is healthy or not. The session tells you nothing. If the seller charges the car to 100 percent the night before because it looks better on the dash, your one measurable test has been erased before you arrive.

So the sentence to send in advance is specific: please do not charge it above about 40 percent before I get there. Most private sellers say yes. Dealerships vary, and a dealership that will not leave a car below half is telling you something about how the conversation is going to go.

Two other conditions travel with that reading, and both need writing down rather than arguing about later. Ambient temperature, because a cold-soaked pack accepts far less power than a warm one and the gap between a cold session and a warm one is large enough to look like a fault when it is not. And whether the car had been driven for half an hour beforehand, since some cars condition the pack on the way to a navigated charging stop and will not do it if you simply park and plug in.

The five minutes at the charger that are worth the whole visit

Pick the charger before the appointment. Not the nearest one: one you have used, on a network whose app is already on your phone with a payment method that works. A test that fails because of a broken stall or a declined card proves nothing about the car and burns forty minutes.

Then, with the car plugged in, watch its own charging screen rather than the pedestal.

Note the peak. Whatever kilowatt figure the car settles at in the first two or three minutes is the number to record, along with the state of charge it was at when it got there. Look up the manufacturer's published peak for that model beforehand, on the maker's own specification page. A published peak is a best case at an ideal state of charge and temperature, so seeing less is normal and seeing a small fraction of it on a mild day is not.

Watch the shape, not just the height. Power that climbs for a minute, holds, and then steps down in stages as the percentage rises is an ordinary curve. Power that starts low and stays flat, or that falls away within the first minute and does not recover, is worth a second session at a different stall before you conclude anything about the car.

Check the pack temperature display if the car has one. Several brands show a battery temperature or a conditioning icon during a session. A car that never leaves its cold-limited state on a warm afternoon is a car to ask about.

Watch what the app charges you. You are paying for this session, and on some networks the meter keeps running after the car stops drawing power. Fees for occupying a stall past the end of a session are the most common surprise on a used-EV test drive, and they are billed by the minute. Unplug when you are done looking rather than when the conversation ends.

Ten minutes of that is enough. You are not filling the battery; you are watching how the car behaves when asked for power.

What a disappointing session actually proves

Less than you will want it to, most of the time. The honest verdict after ten minutes at one stall is usually "nothing conclusive", and knowing which ordinary explanation applies is what stops you either walking away from a good car or buying a bad one on a shrug.

Four things suppress power on a perfectly healthy pack, and all four can be ruled in or out while you are standing there. The state of charge was higher than you were told, because the car got topped up on the way over. The pack was cold, which the car will often say itself if it shows a battery temperature or a conditioning icon. The cabinet was shared: many units split their output between two connectors, so a car on the second one gets a fraction of the headline figure, and the pedestal usually says so on screen. Or the site was derated, which happens in heat and gives itself away because every car there is behaving the same way. Look at what the vehicle in the next bay is pulling before you conclude anything about this one.

Rule all four out and the picture sharpens, but it still does not become a capacity figure. A low, flat curve at a low state of charge on a mild day, at a stall that was feeding the car beside you properly, is a reason to want a second session on another day at another site before committing — and a reason to price that second trip into what this particular car is worth to you. What it is not is a number you can put in an email to the seller. Charging behaviour and measured capacity are two different quantities, and the second one comes out of the car's own diagnostics rather than off a charging screen.

The onboard charger and the 12-volt, while it is standing there

The fast-charge session is the headline, but the car is parked for twenty minutes regardless, and two other things are worth spending that time on. Neither is about the traction pack.

Ask what it does on AC, and hold the answer against the plug you already own. Fast charging is the dramatic test; the onboard charger is the part you will actually use six nights a week, and it is separate hardware from the pack. Trim levels of the same model have shipped with different onboard chargers, so the rate is a property of the specific car rather than of the badge, and the difference decides whether the pack is full by morning or still going at breakfast.

Two things to collect, then. The rated AC figure from the specification page for that exact model year and trim, and the rate the car itself reports on its charging screen the next time it sees a Level 2 unit. Where those two disagree the answer is usually the wall box or the circuit behind it rather than the car — why a bigger wall box so often changes nothing is the same arithmetic seen from the garage end. None of it can be measured at a DC stall, so it stays a question and a lookup. Ask it anyway. It decides whether the charger already screwed to your wall suits the car you are about to buy.

Turn the car on cold, at the start of the visit rather than the end. An EV still runs a small 12-volt battery, and it is what powers the door handles, the contactors and the screens before the traction pack is permitted to do anything at all. When it goes, the symptom is not a slow crank: it is a car that will not answer the key, sitting on a full traction battery, usually on the morning you needed it. It is a consumable and it has nothing to do with pack health, but a car that has spent three weeks on a forecourt being woken for thirty seconds at a time has had a harder life than the odometer suggests. You cannot test it properly in a car park. What you can do is decline to be handed a car that is already unlocked, running and warm when you walk up. Ask for it to be left locked and untouched, and open it yourself.

Two VIN lookups, and they do not answer the same question

Both take a phone and about four minutes. Do them while the car charges.

The federal one first. NHTSA's public recall page at nhtsa.gov/recalls takes a 17-character VIN and reports open safety recalls against that specific car. The limits of the tool matter as much as its output. Every record in the dataset behind it carries an NHTSA campaign number in the agency's own format — 25V130000, 21V560000 — and a manufacturer's customer satisfaction programme or service campaign has no such number and does not appear. Nor is a blank result unambiguous: it can mean the recall was completed, and it can mean there was never one.

Before you go, it is worth seeing the whole campaign list for the model rather than the open items for one car, and NHTSA's public API returns it without any VIN at all:

https://api.nhtsa.gov/recalls/recallsByVehicle?make=ford&model=MUSTANG%20MACH%20E&modelYear=2021

Paste that into a browser address bar and plain text comes back. On 8 September 2026 it returned 13 campaigns for the 2021 Mustang Mach-E, three of them against the traction battery: 22V412000 (reported 10 June 2022), 23V687000 (13 October 2023) and 25V130000 (28 February 2025), all three filed under the component string ELECTRICAL SYSTEM:PROPULSION SYSTEM:TRACTION BATTERY:MANAGEMENT SYSTEM/ENERGY CONTROL MODULE (BMS/BECM). A fourth, 22V333000, carries overTheAirUpdate: true, which means the remedy went out as a software update rather than a shop visit, so there may be no repair order to show you for that one even though the work was done.

The model string is the part that will waste your afternoon. It has to match how the recall database spells the car, and that is not always how anyone else spells it. The Mach-E is stored as MUSTANG MACH E, with a space where the hyphen belongs. Send the hyphenated version and the endpoint answers {"Count":0,"Message":"Results returned successfully","results":[]} — an empty list that reads exactly like "no recalls on this model" and means nothing of the sort. The HTTP status on that reply is 400 rather than 200, but a browser will not show you that.

Looking the name up first has a trap of its own. NHTSA publishes the model list at api.nhtsa.gov/products/vehicle/models?modelYear=2021&make=ford&issueType=r, and that list spells it MUSTANG MACH-E, hyphen included: the spelling the recall endpoint rejects. The two endpoints do not agree with each other. So when a query comes back empty on a car you have some reason to think has been recalled, strip the punctuation and run it again before you believe the zero. The same care applies to the rest of the name — Chevrolet's is BOLT EV, and plain BOLT returns nothing.

NHTSA's use policy asks that the API not be driven with bulk VIN lookups, which is no constraint here. This is one car, looked at twice.

Knowing the campaign list before you arrive changes what you can ask. "Are there any open recalls" is answered with a shrug. "Has 25V130000 been done on this VIN, and can I see the repair order" is not.

Then the manufacturer's own lookup, on the brand's owner site, with the same VIN. This is where service campaigns, field actions and software updates appear alongside the safety recalls, and where a dealer's repair history sometimes shows. Between the two lookups you have a fair picture of what has been done to the car and by whom.

The flag that tells you to stop reading and start asking

Buried in each NHTSA campaign record are three booleans, and they are the closest thing the dataset has to a severity marker: parkIt, parkOutSide and overTheAirUpdate. The first two mean the agency and the manufacturer told owners not to park the car indoors, or not to drive it, pending repair. On combustion cars those usually concern fires from a specific component. On EVs they concern the pack.

The 2019 Chevrolet Bolt EV is the example everyone in this niche already half-remembers. Query that model year and six campaigns come back, of which two are traction battery recalls carrying parkOutSide: true: 20V701000, reported 13 November 2020, and 21V560000, reported 23 July 2021 (api.nhtsa.gov, read 8 September 2026). A used Bolt of that vintage is a perfectly reasonable car to buy, and many of them have had the pack replaced under those campaigns, which is a genuinely good thing to be buying. But "has the module replacement been done, and on what date" stops being a nice-to-know and becomes the question the price depends on.

The general rule that falls out of this: any campaign against the traction battery, on any brand, gets its own question and its own paper. Ask for the repair order, not a verbal yes. A pack replacement also usually resets the clock on part of the coverage, which is a reason to want the document rather than the reassurance.

What the warranty has left is a date, not a model year

The single most misread figure in a used EV listing is the warranty. A 2022 car does not have "four years left of an eight-year battery warranty" unless it happened to be delivered on 1 January 2022. The clock starts on the in-service date, the day the car was delivered to its first retail owner, and on a car that sat as a demonstrator or a press vehicle that date can be a long way from the model year.

Getting it is a phone call. Any franchise dealer's service department for that brand can look up a VIN and tell you the in-service date and what coverage remains. Ask for it in an email rather than over the phone, and ask for the coverage line by line, because the battery, the drive unit and the rest of the powertrain frequently run on different terms.

Two further things are worth having in the same email, because they are cheap to ask for while someone already has the VIN on screen and expensive to establish afterwards: whether the battery coverage is recorded against this car as transferring to you, and whether any pack or module work has been carried out under warranty. The second one is the paper trail for any traction-battery campaign you turned up above, coming from the manufacturer's system rather than from the seller.

Do not take the transfer answer from the brand's current marketing page. Coverage terms are set by the booklet issued with that model year, some of those booklets do not say what their own summary charts appear to say, and what the threshold and the transfer clause actually read like on the page is a twenty-minute job worth doing before the visit rather than after the deposit.

Open the boot well before you sign, not after

This is the least technical check and the one most often skipped, and every missing item is real money.

The adapter deserves a moment on its own, because which plug goes with which network changed twice while these cars were being sold, and the adapter in the frunk is frequently the wrong one, pointing the wrong way. An adapter that lets a Tesla use a CCS station is a different object from one that lets a non-Tesla use a NACS station, and owning the first does not help you at all if you need the second. Which adapter your particular car needs, and what it cannot do for you is worth settling before the visit so you can look at the actual object in the boot and know whether it is the one you need.

Ask for photographs of all of it before you drive over. An item that appears in a photograph on Tuesday and not in the boot on Saturday is a conversation you can still have.

Turn the visit into a record, because in two years you will need one

Everything above produces numbers, and every one of them was taken on a car you do not own, on a day you will not get back. Four minutes at the end fixes that, and the method is dull: photograph rather than remember.

Take a picture of the odometer, the VIN plate, the state-of-charge display before the charging session and after it, the peak kilowatt figure on the charging screen, and any battery health or capacity screen the car has. Then, in a note on your phone, write the date, the ambient temperature, whether the car had been driven beforehand, and the name and location of the charging station.

That context is what makes the visit worth anything twice over. Before you buy, it is the only version of the day you can put in front of the seller: a peak figure with a state of charge, a temperature and a station name attached is a position, while "it seemed slow" is a feeling. After you buy, the same lines become entry zero of the log the car deserves from then on, taken before you had any reason to flatter it. What to record every six months once it is yours, and from which source is a slightly different list from this one, and it only works if the first entry exists.

One thing worth knowing before you hand over money: the charging history lives in the seller's account, not in the car. Once the vehicle is transferred out of their app, the record of how often it was fast charged goes with them. If the car's own history matters to you, and on a high-mileage example it should, ask the seller to open their app and scroll it while you are standing there. Some cars are more forthcoming than others. The Nissan LEAF keeps explicit counters for quick charges and for Level 1 and Level 2 connections, readable through third-party diagnostic apps, and a car with a very high quick-charge count against modest mileage has had a working life rather than a commuting one.

The order that fits in ninety minutes

Nothing in that sequence requires a scan tool, a mechanic or a specialist inspection, and none of it depends on the seller's goodwill beyond leaving the car partly discharged. What it produces is four pieces of paper: a charging measurement with conditions attached, a recall status per campaign number, an in-service date, and a photographed inventory of what came with the car.

If you get to the end of the visit with only three of them, the one to go back for is the in-service date. It is the cheapest to obtain and the one that changes the price.