When to Replace a Bike Chain

Heads up: we earn a commission if you buy through the links on this page, at no extra cost to you. It never decides what makes the list — here’s how that works.

A chain is one of the cheapest parts on your bike. The cassette it drives is not, and the chainrings are worse. Replace the chain on time and it stays that way. Miss the window and the chain takes the cassette with it, which turns a small routine purchase into a large annoyed one.

That is the entire argument for owning a chain checker, and Park Tool — which sells them, and also publishes the guidance for free — puts the cost case plainly: “Since it’s far more expensive to replace your cassette than it is to replace a chain, knowing when to replace your chain can actually save you some money in the long run.”

So: the numbers, where they come from, how to measure, and what happens if you don’t.

Chains don’t stretch

The word “stretch” is everywhere in this topic and it’s wrong, in a way that matters for understanding the rest.

The steel side plates are not getting longer. What’s happening is that the pins and the insides of the rollers are wearing away. Each link develops a tiny amount of slop, and there are a hundred-odd links, so the slop adds up into a chain that’s measurably longer end to end. Park Tool describes it exactly this way: “The internal parts of the chain (the rivets and rollers) begin to wear out, giving the illusion of stretching.”

Why it matters: this is why a worn chain destroys a cassette.A chain is supposed to have its rollers land at a fixed pitch that matches the spacing of the cassette teeth. As the pins wear, that pitch grows, and the rollers stop sitting in the bottom of each tooth valley — they start climbing the tooth faces instead. Now every pedal stroke is filing the teeth into a new, longer shape to match the worn chain. Which means a fresh chain fitted to that cassette won’t mesh either: you get a chain that skips under power, and now you need both. The chain does the damage slowly, over the last fraction of a percent, and then presents the bill all at once.

The number everyone quotes — 0.5%, 0.75% — is that pitch growth, expressed as a percentage. Half a percent sounds like nothing. Across a 12in span of chain it’s about a sixteenth of an inch, and it’s the difference between replacing one part and three.

The numbers, from Park Tool

We are not going to state these from memory, and you shouldn’t take them from anyone who does. Here is what Park Tool publishes, retrieved on 17 July 2026 — a company with no stake in which brand of chain you buy:

• Single-speed: replace at 1.0%
• 10-speed cassette or freewheel, or fewer: replace at 0.75%
• 11-speed cassette or more: replace at 0.5%

Park Tool’s instruction is unambiguous: “Your chain should be replaced by the time it reaches the wear value specified.” By the time it reaches it — not once it’s comfortably past.

The pattern is easy to remember once you see the logic. More speeds means a narrower chain and narrower cogs, which means less metal, which means less margin for error. An 11 or 12-speed drivetrain is a more precise machine and it tolerates less. It’s also, inconveniently, a more expensive one to replace — so the drivetrains with the tightest threshold are exactly the ones where missing it hurts most. If you’re not sure what you’re running, our groupset hierarchy guide will tell you.

How to measure it

The chain checker (10 seconds, and the one to use)

A go/no-go gauge. Park Tool describes the CC-3.2 as “designed to accurately indicate when a chain reaches 0.5% and 0.75% wear (or ‘stretch’)” and as “a go/no-go gauge” that “works on most derailleur chains from 5 to 12-speed”.

It has two ends, marked 0.5 and 0.75. You hook one end into the chain and try to drop the other in. If it drops in, you’re past that number.That’s the whole tool. Match the end you care about to the table above: 11-speed and up, use the 0.5 end and replace when it seats.

The important caveat, published by Park Tool itself: the CC-3.2 is not the right gauge for SRAM Flattop and T-Type 12-speed chains — those need the CC-4.2. Twelve-speed is where the exceptions live generally, so check what your chain maker says before trusting any gauge on it.

The ruler (free, and fine)

You don’t have to buy anything. Park Tool publishes the ruler method too:

“Pick a rivet and line it up at the zero mark. Count 24 more rivets and your last rivet should be at the 12″ mark of your ruler. If it is off by more than 1/16″ your chain is stretched to the point of replacement.”

A sixteenth of an inch over twelve inches is roughly 0.5% — which is the 11-speed threshold, so this method is conservative for 10-speed and correct for 11. It is fiddlier than a gauge and much easier to get wrong by a rivet, but it costs nothing and it’s the same underlying measurement. If you own a ruler and you’re not sure you’ll ever measure a chain twice, start here.

What doesn’t work

Lifting the chain off the front chainring to see how far it comes away is a folk test that tells you very little, and “it still shifts fine” is not a measurement — a chain can be well past 0.5% and shift perfectly, right up until the new one skips. Measure it. The whole point of the threshold is that it arrives before you notice anything.

What ignoring it costs

Roughly, and the arithmetic is yours to do with live prices rather than ours to invent:

• Caught on time: you buy a chain. See the chain roundup for live prices.
• Caught late: you buy a chain and a cassette, because the worn chain has reshaped the cog teeth and a new chain will skip on them.
• Caught very late: chain, cassette and chainrings. The rings are the most expensive of the three and the last to go, which is the only mercy in this.

Park Tool’s framing again: “Using a chain beyond its intended wear limit will prematurely wear out your cogs and chainrings, so staying on top of this routine maintenance task can save you a lot of cost and hassle.”

Put next to a chain checker — the cheapest tool in this entire hub, check the live price above — the case makes itself. This is the rare purchase where a single correctly-timed replacement pays for the tool several times over, and you’ll own it for a decade.

How often should you check?

Here is where we stop being able to give you a number, and we’d rather say so than make one up.

Chain wear rate depends on your lube, your conditions, how clean you keep the drivetrain, your power, and how much of your riding is in the dirt. The spread between a waxed chain on dry tarmac and a neglected one through a wet winter is enormous — and we have not measured either, so any mileage interval we published would be fiction with a decimal point in it.

The most extensive public data on how fast chains wear under different lubricants comes from Zero Friction Cycling, which states it has run over 300,000km of controlled testing and publishes the results. ZFC also sells chains and lubricants, including products that test well in its own results — its site runs a store and states it stocks “only genuinely proven top products”. That is a structural commercial interest in its own findings, and every publisher citing ZFC as “independent” lab data — which is most of them — is describing it inaccurately. We link it so you can read the raw data and weigh it yourself rather than taking our word or theirs.

Our honest answer, with no data behind it and labelled as such: check it whenever you clean the drivetrain properly. That ties a ten-second measurement to a habit you already have, which beats any interval you’ll forget. The cleaning procedurestarts with this check for exactly that reason — Park Tool puts it at step one of its own method, because detailing a chain that’s already finished is a waste of your afternoon.