Research · Chain wear
When should you replace it?
Your chain never stretched. The pins and rollers wore, the spacing opened up — and the part that gets expensive if you ignore it is the cassette, not the chain.
Every drivetrain has a number at which its chain should come off: 0.5% elongation on 11-speed and up, 0.75% below that. Ride past it and the worn chain starts reshaping the cogs it runs on. The rule is simple and well documented. The awkward part is that the cheap gauge most riders measure it with may be the least trustworthy component in the process.
Reviewed July 2026 · 4 sources
- Chains do not stretch. Park Tool is explicit that the side plates hold their length and it is the internal parts that go: “The internal parts of the chain (the rivets and rollers) begin to wear out, giving the illusion of stretching.” What you measure is accumulated internal clearance, expressed as a percentage of the chain's original pitch.
- The replacement threshold is set by how many sprockets the chain has to sit between. Park Tool publishes three: 0.5% for an 11-speed cassette or more, 0.75% for 10-speed or less, and 1.0% for single-speed. Narrower modern chains get the tightest limit.
- The cost of ignoring it lands on the other parts, not the chain: “Using a chain beyond its intended wear limit will prematurely wear out your cogs and chainrings.” A worn chain meshes badly and reshapes what it runs on, which is why the chain is the cheap consumable you replace to protect the expensive one.
- Wear rate is driven far more by what's on the chain than by how far you ride it. In Zero Friction Cycling's lab testing, reported by CYCLINGABOUT, the best drip wax wore 0.12 chains per 5,000km in dry off-road conditions against 1.76 for a typical wet lube — roughly a 15× spread between two bottles on the same shelf.
- The shop-floor version of the rule — “replace the chain to save the cassette” — is repeated widely enough to be near-universal advice, and it follows directly from the manufacturer guidance above, but the break-even point for any particular rider depends on their own parts prices and mileage.
- Riders running immersion-waxed chains report drivetrain lives far beyond the usual expectation; CYCLINGABOUT notes waxed chains “have been known to go for 15,000 to 20,000 km.” Treat that as the top of the reported range under a demanding maintenance routine, not a number to plan around.
- Zero Friction Cycling's testing notes that cheap chains “will often rip past 0.5% wear in a blink” and take the cassette with them — a useful warning, though the phrasing is theirs and the wear curve isn't published per-model.
- The measurement is the least reliable part of the process. Zero Friction Cycling reports that chain checkers disagree badly with one another — “a common problem is accuracy from one tool to another. Many are cast and finished or simply cut to very rough tolerances” — and gives a concrete case of a customer's tool “reading 0.75, only to find when checked with digital checker or shimano Tl-CN-42 that it is only around 0.25.” A 0.5% rule is only as good as the gauge, and the gap they describe spans the entire decision.
- Park Tool allows the 12-inch ruler method as “a good starting point if you haven't replaced your chain in a while” while stating it is “not as accurate as using the proper tool” — but doesn't quantify the error, so how much you're giving up by using a ruler is unpublished.
- Why the cut-off sits exactly at 0.5% rather than 0.4% or 0.6% is manufacturer guidance rather than a published experiment. The direction is well established; the precise number is a convention, and the thresholds differ by speed count because of sprocket spacing rather than because each was independently derived.
What the percentage actually measures
A new chain has a fixed pitch — the distance from one pin to the next. As it runs, the pins and the surfaces they turn against wear away, and each link gains a fraction of a millimetre of slack. Multiply that by a hundred-odd links and the chain reaches further around the cassette than it used to. Nothing has stretched in the material sense; Park Tool's phrasing is that the wear gives “the illusion of stretching.” The percentage you read off a checker is that accumulated slack expressed against the original pitch, which is why 0.5% sounds tiny and still matters.
It matters because the chain and the cogs have to mesh tooth-for-roller. Once the spacing has opened up, the load stops sharing evenly across the engaged teeth and concentrates on fewer of them, which wears the cog into a profile that then only accepts a worn chain. That's the trap: fit a fresh chain to a cassette that has been reshaped by an old one and it skips under power, so the bill arrives for both. Replacing on schedule is the cheap move precisely because it keeps the expensive part in spec.
Why the tool is the weak link
The rule is easy; trusting your reading is not. Zero Friction Cycling — a lab that does nothing but test chains — reports that checkers vary enough between units to change the answer, and describes a customer tool showing 0.75% on a chain that measured around 0.25% on a Shimano TL-CN-42 and a digital gauge. That isn't a rounding difference. One reading says throw the chain away immediately; the other says it's barely run in.
The practical response is not to buy a more expensive gauge but to be consistent and sceptical. Measure the same chain in several places rather than one, since wear is rarely even along its length, and treat a single alarming reading as a prompt to check again with something else before you spend money. ZFC's own recommendation is to “replace chain just before or at 0.5% wear when any section of chain is reaching this point” — note *any section*, not the average. If you only ever own one tool, knowing its bias matters more than knowing its price.
Sources
- When to Replace a Worn ChainPark ToolThe thresholds — 0.5% for 11-speed+, 0.75% for 10-speed or less, 1.0% single-speed — plus the mechanism (“the rivets and rollers begin to wear out, giving the illusion of stretching”) and the consequence (“will prematurely wear out your cogs and chainrings”).
- Shimano Chain Wear CheckerZero Friction CyclingChain checkers disagree materially: a customer tool read 0.75% where a digital gauge and the Shimano TL-CN-42 read ~0.25%. Recommends replacing “just before or at 0.5% wear when any section of chain is reaching this point.”
- The 10 Best Bike Chain Lubes According To Science (Lab Tests)CYCLINGABOUT, reporting Zero Friction Cycling test dataLubricant choice dominates wear rate: 0.12 chains worn per 5,000km (best drip wax) vs 1.76 (typical wet lube) in dry off-road conditions — ~15×. Also the reported 15,000–20,000km figure for waxed chains.
- Chain TestingZero Friction CyclingCheap chains “will often rip past 0.5% wear in a blink,” taking the cassette with them — the economic case for monitoring wear rather than running to failure.