A used electric bike gets looked over like an ordinary bike nine times out of ten: you squeeze the tyres, spin the wheels, check the frame is straight, haggle over a tired chain. And you miss the point entirely, because on an e-bike the battery and the motor easily account for half to two thirds of what it would cost to replace the machine.
Neither of those two parts can be judged by eye. A pack that has lost 35 % of its capacity looks exactly like a new one, and shows the same five bars on the display. A motor with bearings on the way out makes a noise that passes for normal if you have never heard the same motor in good health. Hence these seven checks, in order: the first three decide whether you buy, the next four decide the price.
1. The pack: remaining capacity, not the age on the advert
A lithium cell ages in two ways. Through cycles, meaning the number of full charges it has accumulated: reckon on 500 to 1000 cycles before it drops to somewhere around 70 to 80 % of its original capacity. And through the calendar, simply by existing: a loss of the order of 2 to 5 % a year is added on top, even on a bike that has not been ridden, and it speeds up sharply if the pack has spent its winters sitting at 100 % in a warm garage.
The counter-intuitive consequence: a three-year-old pack with 400 cycles and sensible storage is often in better shape than a six-year-old pack with 120 cycles left full in a garden shed. "Battery barely used" is therefore not an argument, it is an incomplete piece of information.
On the day: charge it fully, note the range shown in the strongest assistance mode, then ride a loop you know. A healthy pack returns a range consistent with its capacity in watt-hours. If it drops three bars in five kilometres, or if the display cuts out abruptly at a third of charge, a cell is out of balance: the bike is not worth that money.
2. Cycles and motor mileage, read out of the electronics
Most branded drive systems (Bosch, Shimano, Yamaha, Brose and Bafang among others) log the real mileage, the number of pack cycles and the error code history in their controller. That memory does not reset when you swap the display, whatever plenty of sellers believe.
An authorised dealer can plug in their tool and print that report, often for a modest fee. On an e-bike advertised at more than £800, it is worth the trip: it is the only document that tells "4000 km" apart from "4000 km showing on a replacement display". If the seller refuses to go, you have your answer.
Look at the stored fault codes too. A speed sensor fault coming back regularly means a magnet or a connector: a few tens of pounds. A recurring motor fault is a different budget altogether. The brand-by-brand decoding is set out in the guide to electric bike error codes.
3. The motor: noises, play and heat
With the bike on a stand or lifted, turn the cranks by hand with the assistance off. You should feel even resistance and hear at worst a faint whine of gears. What should worry you: a cyclic tight spot (a notchy bearing), a dry rattle under load (damaged gear train), a noise that changes when you push sideways on the crank arm.
Then check for play at the bottom bracket: hand on a crank arm, pull and push at right angles to the axle. Perceptible play on a mid-drive means the axle bearings, and the job is done with the motor out. Finally, after your test loop, put a hand on the casing: warm is normal, too hot to hold after a flat ride is not.
4. The drivetrain: it wears two to three times faster
A mid-drive motor routinely puts 50 to 85 Nm through the chain. As a result, an e-bike drivetrain often needs replacing every 1000 to 2500 km, where an ordinary bike will run 3000 to 5000 km on the same chain.
Take your chain wear gauge along, it fits in a pocket. Beyond 0.5 % elongation on a modern 11 or 12-speed drivetrain the chain is due for replacement; beyond 0.75 %, the cassette and chainring have gone with it. Look at the shape of the chainring teeth: if they have taken on an asymmetric hooked profile, the whole lot is on the bill. A mid-drive-specific chainring costs more than a standard one, and is not always interchangeable.
