Shimano cranks failing. Time to do something about it?

I’m sure many of us on here are aware of the high failure rate of Shimano hollow tech cranks. But Shimano still seem to be claiming that it isn’t an issue. Is it time for CT nerds to get Shimano to fess up to their poor design and do something about it?

I alone know 2 friends in my club that have had their cranks split. Multiply that globally it must be massive. So big that there’s a dedicated Meem account focusing on the issue.

Why are they failing, why hasn’t the big S fussed up to it, why haven’t they changed design? Nerds assemble!


It is interesting, it also appears many cycling websites are strangely not interested in reporting on it? Maybe being “cancelled” by Shimano is the fear factor?


TWO!! I had two different ultegra cranks break on me. One snapped while riding, one I caught beforehand. Total bs.

1 Like

I work at a bike shop, and I would estimate that I have come across somewhere in the range of 400-500 of these cranksets. Of all those, I have seen 3 failures, so a little under 1%. Luckily, no one was seriously injured in any of the failures - only one was catastrophic (snapped off under power) while the other two were caught during service shortly after the delamination had started. Shimano warrantied all three, even though two were outside the warranty window.

This is in Atlanta GA where humidity is quite high in the spring and summer.


Peak Torque on YT did a good analysis of this and postulates it may be to do with higher humidity levels

While I haven’t (yet) heard of any failures in the groups I ride with in usually-low humidity Melbs, I was sufficiently paranoid about a failure and (in these days of parts shortages and lengthy shipping delays) being stuck with no cranks to buy a spare set just in case.

1 Like

I’ve not heard of any failures first hand, or experienced any myself. I lived in Singapore (high humidity, anyone?) for nine years (2011-2020) and ran hollowtech Ultegra 6700, Dura Ace 7900 and 9000 in that time.
No issues, and haven’t heard first-hand of any.


Lol… what exactly do you want them to “do” about it?

I’ve personally seen hundreds of these cranks; the 6800, 8000, 9000, and 9100 cranks… and have owned six sets myself of 9000 and 9100 cranks. I have seen exactly zero failures. Interestingly, in that same time, I’ve snapped an X01 crankset, bent an XT crankarm, and blown the pedal insert clear out of an XX1 crankarm. This is a non issue, there’s so many of these cranks out there that the failure rate is statistically irrelevant and just being wildly overhyped.


I was also wondering how prevalent this issue is after seeing a few posts on social media, but it seems to me that extrapolation of anecdotal evidence isn’t quite so reliable after all…

1 Like

I know that Time make good frames and pedals but why should they have to sort out Shimano’s problem? :wink:


So you haven’t personally seen it so that means it doesn’t happen or worth acknowledging?

If I was to have a serious accident as a result of one of these cranksets failing in the same way I’d be pretty unhappy, obviously the numbers are not massive within the total production but there’s enough to be some sort of issue.

A total recall required, obviously not but I find it interesting that there are recalls from other manufacturers with only one or two failures.


Always remember: the plural of “anecdote” is not “data”.


I have seen in the flesh one broken DA 9000 (outside half of the right arm decided it had had enough at the pedal insert). The rider was in for a good fear (he barely managed to stay upright) and a long one-leg ride home.

1 Like

When you mention Time in a crankset failure topic, I can’t help but think of this:

There are my own Stronglight Helion (was a very fine crankset, light and stiff with a titanium hollow axle and a high quality bottom bracket). The axle eventually turned in the right crank arm (though it was the revised version, revised because the first ones all ended up like this).

I mention Time because those crankset were also branded as Time (ASX was the name):


I’m in Melbourne and had the Ultegra crankset on my 2016 Defy replaced under warranty in March. I took it to my LBS for a service for what I thought was a creaky BB, but which turned out to be the infamous Ultegra drivetrain failure. So it does happen here, but I presume not often as the mechanics at the LBS had never seen it before.

1 Like

Just to be clear, your devalue the anecdotal experience of myself and literally thousands of other people and shops who’ve seen hundreds of cranks but place significant value on the anecdotal experience of a tiny but vocal minority who has had failures?

That’s logically consistent and totally makes sense.


Lol… what exactly do you want them to “do” about it?

Acknowledge the problem and offer some sort of extended warranty for failures.

This is a non issue, there’s so many of these cranks out there that the failure rate is statistically irrelevant and just being wildly overhyped.

Unfortunately, we all tend to have a bit of a status quo bias, although some more than others. This can be fine, but if there’s a genuine problem, it obviously is not fine. It’s true that the plural of anecdote isn’t data. However, at some point you can’t ignore all the anecdotes. I think we should be able to agree that Shimano cranks are failing at a higher rate than is acceptable. Shimano’s Hollowtech process isn’t truly pushing the bounds of low weight and it should be a mature process overall. Nobody is saying that the failure rate is 10-20%. It may not even be 1%. However, a crank shouldn’t disbond mid-ride like that.

I’m sort of reminded of Apple’s response to the infamous butterfly keyboards. A lot of them failed. It was true that overall, Apple’s designs have been reliable, and I don’t know what the actual failure rate for the butterfly keyboards was (I will guess that it was higher than Shimano cranks, though). Whatever the case, the first commenters on the Internet met similar responses as the one quoted. Heck, even in 2017-2018, when it was well apparent that there was a problem, you’d still get responses like this.

1 Like

Don’t start me on Apple’s latest designs, or I’ll show the flexgated Macbook Pro I now use as a twice-as-more-expensive-and-big Mac Mini, only because the 2016 MBP is covered by the recall and the 2017 not (though they are the exact same machines).

1 Like

The problem is that no one here has any accurate idea of how widespread the problem is: even someone who works in a large bike shop will only have (more) anecdata. People also tend not to make IG posts along the lines of ‘Hey, my 105 cranks didn’t break today!’, so there’s a selection bias to the information we have.

Having said that, there’s clearly an established failure pattern, and it is potentially dangerous. What’s unknown is how common it is.


Guys, just thought you should know that my Shimano 105 cranks didn’t break today.



They did acknowledge the problem…

“Thank you for bringing the reported failures of cranks to our attention. We take customer complaints seriously and, as a result, we initiated an investigation.
“Crank failures do occur, even though our cranks do not have any design problems or if there are no other easily identifiable issues.
“During our examination of usage cases, and through our own internal testing, we have not identified a design problem with the cranks, and we are continuing our investigation to discover other factors or causes.
“We would like to be able to give further details, but we cannot at this point because our investigation has not resulted in the identification of a cause of the crank failures your readers reported.
“We would like you to know that your readers’ complaints are not being ignored, and Shimano’s engineers are trying to find out if there is a specific cause.”

Why would they warranty an issue that they have not been able to identify as being resultant from their manufacturing or materials?

I have no bias for the status quo… I have a bias for reality. And the reality is that there are likely millions of these cranks out there at this point, with what… perhaps hundreds of failures, maybe a few thousand? So, we’re not talking about 1% failure rate we’re talking about a fractional percent failure rate. Maybe a 0.1% failure rate, or a 99.9% success rate… That’s absolutely an acceptable product failure rate, particularly when the majority of anecdotal reports come from older models which are no longer even in production and from riders who are well beyond the warranty period.

Without some quantifiable study conducted that isolates the riding and maintenance conditions over the life of the product and identifies a real issue, I think Shimano’s response has been just fine.