When to replace a bike chain, and why 0.5% is not one number

Chain life ranges from a few hundred kilometres to well over ten thousand depending on lubricant, conditions and how the chain is cleaned. That is not a spread you can average. Any figure of the form 'replace every N km' is a guess dressed as a spec.

What manufacturers and tool makers actually specify is elongation: how much longer the chain has become than it was new, as a percentage. That is a measurement you take, not a number you accumulate.

Park Tool publishes 0.5% for 11-speed and above and 0.75% for 10-speed and below. Those are the figures nearly everyone means when they quote a chain limit.

SRAM is the exception worth knowing: SRAM support gives 0.8% for Eagle and ships its own gauge for it. Replacing an Eagle chain at 0.5% means replacing it roughly 60% more often than SRAM asks, which is real money on a 12-speed chain.

Shimano publishes no elongation percentage at all. The universally quoted 'Shimano 0.5% / 0.75%' comes from the TL-CN42 gauge, which is a 7-to-11-speed tool – not a spec sheet. Attribute those numbers to Park Tool or to industry practice, never to Shimano.

A Park CC-2 or similar caliper-style tool gives a continuous reading you can trend over time. A CC-3.2 or CC-4 is a go/no-go gauge: it tells you which side of a threshold you are on and nothing more.

Mixing readings from the two kinds silently corrupts any trend line, which is why Kuralta records which tool a measurement came from alongside the number.

A chain replaced on time wears only itself. A chain ridden past its limit reshapes the cassette and the chainrings to match, so the next new chain skips under load and the drivetrain has to be replaced together.

That is the entire economic argument for measuring: a chain costs a fraction of a cassette, and measuring takes under a minute.

Lubricant. Independent chain testing by Zero Friction Cycling found more than a fifty-fold difference in wear rate between the best and worst lubricants under the same conditions – a far larger effect than anything else you control.

Then conditions. The same testing puts a wet ride at roughly 2.4 times the wear of a dry one, with a lasting penalty on the block after it until the chain is stripped and re-treated. Those are single-lab figures rather than physics, and they should be treated as tunable defaults, but the direction is not in doubt.