I measured the wear on a new Shimano 10-speed HG-95 chain by measuring the length of a certain number of links with a caliper like this (the jaws of the caliper are inside the chain): Caliper beside chain, dial reading about 2.8 But the results don't make any sense to me.

When measuring 11 links I got a length of 132.7 millimetres which should be equivalent to a length of 120.6 mm on the more common 10 links used as a reference. Obviously this is way too long, as a new chain should be 119.5 millimetres long.

I got another caliper and measured 7 and 8 links (because it isn't long enough for more links) and got a length of 116.9 and 118 millimetres which is too short.

I switched back to the first caliper and got decent results when measuring 9 links (about 119.2 mm) but now I am very confused and don't know what I am doing wrong, because as far as I know my calipers are at least decently accurate.

  • 4
    Please post some photos of how you are using the calipers to take the measurement. Commented Aug 2, 2022 at 14:46
  • 1
    @RossMillikan I hope this clears it up Commented Aug 2, 2022 at 18:08

3 Answers 3


First, 10 links is not 119,5 mm. The pins on the chain are half an inch apart. Counting 10 pins would give you 5 inches, or 12,7 cm (1 inch = 25,4 mm).

Where are your measurement reference points? Usually one would measure from the center of one pin to the center of the next reference link. You can use the edges of the links if you find that easier, but I don't find the accuracy of my method to be a problem. When a chain is (sufficiently) worn, you will be able to see it even if you eyeball the reference points.

See the image below (excuse the dirty background). The five links of the brand new lower (thicker, single-speed) chain are almost exactly 5 inches long (lens distortion, camera angle etc. may skew the measurement, but I assure you the chains are aligned and the caliper shows 5" exactly):

enter image description here

Note that the green and brown lines capture 5 inches of the chain, the blue line on top only captures 4 links. Sorry if this is confusing.

NB: The problem I've had is that measuring just a handful of links gives iffy results. Maybe if the chain is tensioned and on the bike, this works well, but laying them flat on a table, I can never get a nice result. I couldn't tell from my own image, nor from the 5 inches of chain right in front of me, that the top (narrower, 10-speed) chain pictured, is used, and warrants replacement. Thankfully, while the relative error stays constant, the absolute error increases with the length of the chain. Pictured below is this same 10-speed chain over a brand new chain, but after 10 inches (I did my best to keep them aligned at the reference link, ten inches to the left):

enter image description here

I personally don't trust cheap chain measuring tools to be able to measure this with any sort of repeatability. What chances are there for a stamped piece of metal to be accurate to within 1% at 127 mm? I compare the chain with a brand new one, across the 114 links supplied, which will show you immediately and without a doubt whether your old chain is stretched. I also don't see the value in spending $20 or more on a more expensive tool given that a new chain will cost me less.

  • I added a picture showing how I measured the length. This method is quite popular at least in the German part of the internet (this is also where I got the 119,5mm for new chains). I have never heard about your methods. How big can the deviation between the two chains get before the chain needs replacement? Commented Aug 2, 2022 at 18:15
  • 1
    Pin and roller wear are two different variables that both contribute to chain wear. Note that juhist's answer of 127 mm implies the same measurement method, i.e. not between the rollers, but between the links. We're saying, unless you have a tool that slots between the rollers, it is easier to measure the distance between the links (which rounds to exactly 25.4 mm for a new chain).
    – jayded-bee
    Commented Aug 2, 2022 at 18:26
  • 1
    Your chosen threshold for chain replacement (usually 0.5% or 0.75%) will stay the same regardless of the measurement method. Again, the relative error (in percentage points) is a constant, but the absolute error (the actual deviation in, say, millimeters) will change. A chain stretch of 0.9525 mm over what used to be 127 mm when new amounts to a relative stretch of 0.75%, but then a stretch of 0.89625 mm over what should be 119.5 mm also turns out to be a 0.75% increase.
    – jayded-bee
    Commented Aug 2, 2022 at 18:26
  • 1
    To be fair, I've read of your chosen method before, but I personally came to the same conclusions you're showing us here: The measurements made no sense, and a normal set of calipers is too small to accurately judge the difference.
    – jayded-bee
    Commented Aug 2, 2022 at 18:29
  • 2
    So I tried out your method on the new chain and instantly got the expected result. I guess I will now use your methods Commented Aug 2, 2022 at 18:34

If you are using calipers, then you need to either eyeball against a reference point, or butt the jaws against points that are not on-center and then measure back to center and subtract the difference. Either method seems likely to introduce inaccuracies. There are dedicated chain-measuring tools (for example) for this purpose.

  • I just added a photo showing how I measured the length, which should be quite accurate from my understanding Commented Aug 2, 2022 at 18:09
  • 1
    @user11914177 That's pretty much what I expected. Based on that photo, you need to subtract half the diameter of the roller to get an accurate measurement for an integer number of links or half-links.
    – Adam Rice
    Commented Aug 2, 2022 at 19:40
  • Measure the chain with the calipers on the rollers, then measure the diameter of the rollers and subtract one roller diameter from the measured length. Commented Aug 10, 2022 at 14:37

Usually when measuring chains by eyesight, you measure 24 half-links or 12 full-links, requiring an inch ruler that has 1/16" marks and is longer than 12 inches. If using an exactly 12 inch long ruler (these are most common), then you have to extrapolate with a worn chain as the ruler ends at 12" and a worn chain can be 12"+1"/16.

Working with millimeters is harder. The problem with calipers is that they usually end at about 14 centimeters or so, and 6 full links would be too long for a caliper so you have to use 5 full links. That's 127mm when unworn and 127.63mm when worn.

A caliper can perfectly well measure to an accuracy of 0.1mm, but only when you measure not by eyesight but by feeling. Chains are measured by eyesight and if measuring by eyesight there easily can be 0.3mm error. This error would be 50% of the difference between new and fully worn.

Measuring halfway between full links is not recommended, as every other half-link wears in a different way, so only full link measurements are accurate.

I have found 24" rulers to be too long to conveniently store and 12" rulers to require extrapolation by eye. Finding 13" rulers which would be optimal for chain measurement is practically impossible. So I use good quality chainwear tools such as Park Tool CC-4, Shimano TL-CN40, Shimano TL-CN41, Shimano TL-CN42 or Pedro's Chain Checker Plus II. Most chainwear tools are crap because they do the measurement at different sides of chain rollers so they measure pin wear and roller clearance at the same time, being sensitive to roller diameter as well. Good tools only measure pin wear, not measuring roller clearance at all (roller clearance doesn't matter here), and are insensitive to roller diameter, because they do the measurement at the same side.

I don't think that with normal eyesight it's possible to correctly identify chain wear state with a millimeter ruler, or a caliper, or any measurement shorter than 12 inches. With 10 inch measurement for example, the 1/16" mark would be 0.625% wear which could be arguably a little too much, although it might be barely acceptable if not desiring maximum cassette life.

  • Thats an interesting method but I never heard of it. I added a picture showing how i measured the length, it is quite common at least in the German part of the internet. Does my method suffer from the same problems as the cheap chain wear tools? Commented Aug 2, 2022 at 18:21
  • Working in metric it's easiest to measure 10 full links: 254mm nominal. 255mm is 0.4% (to 1 significant digit) and 256mm is 0.8%. With 1mm graduations on a typical 30cm steel rule (the 0.5mm marks tend to only be for the first 100 or 150mm) you can easily interpolate to 0.5mm, and 255.5mm (0.6%) is an easy threshold with enough margin for a new chain to arrive. Don't forget that working in mm you've got 1/25" increments
    – Chris H
    Commented Aug 10, 2022 at 19:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.