In the past, I kept using a chain, until one of the sprocket teeth broke. It was a 10 speed, and I had friction shifters. I guess the chain kept riding higher and higher on the surface of the teeth, until the amount of metal at the top of the tooth wasn't enough to withstand the stress. I had replaced the rear derailleur pulleys with bulls-eye pulleys (sealed cartridge bearings). Shifting was great.

Then I learned about chain wear, and I started replacing my chain so I wouldn't wear my sprockets.

However, I think the current 12-1/8" chain replacement is too early. The chain will keep riding higher on the teeth, but, in doing so, it won't wear the lower part of the teeth, where a new chain will engage. Wearing the upper part of the teeth is free, as long as you don't break a tooth. So you can put in a new chain on some really worn sprockets, as long as it's the first time you replace the chain. Sure, there will be ramps on the teeth, to drive the chain up higher, but the ramps will have been made by the first pass of the first chain. I think the new chain won't go anywhere near the wear made by the old chain on it's last month of service.

Ok, I've been talking like I know. It is an educated guess, actually. I'd like to know your experiences, in using a chain well past the 12-1/8" wear point, and what happened when you put in a new chain.

reference: Chain Maintenance

All my experiences with worn sprockets and newer chains is that you get a very annoying drive train. I'd rather just replace both once they get irritating. Its not like a 9 speed chain and sprocket are really expensive.

IME the indicators that 12-1/8" replacement is about right is reasonable and I follow it but if you believe the zillions of miles people have put on chains over the years that gave us that standard is wrong maybe it is just your riding conditions that support your "educated guess".

The chain doesn't just suddenly become worn and start engaging higher on the sprocket teeth and sparing the lower portion from wear, nor does a worn chain ride high on the teeth for the sprocket's full rotation; as the chain engages and disengages the sprocket, it slides down the tooth and back up again. A worn chain wears the whole engaged side of the sprocket tooth.

Worn chains cause worn sprockets. Worn sprockets don't mesh perfectly with new chains. When sprockets' and chains' pitch don't match well enough, you get chain slippage under load way before you get broken teeth.

If that is the case, then you are right. I was going by what I believe, and what is written here:
"a chain that has worn along with the sprocket may ride high on the teeth all the way around. I have yet to check this -- John Allen" Chain Maintenance

- Yes, the time I broke a sprocket tooth, I had never changed the chain.
- Worn chains may just wear the sprocket higher up on the tooth.
- If you keep riding a chain that slips, in a couple hundred miles they will wear together and not slip any more (found someone say this in another thread, but didn't save the link).

I have lived outside for months at a time, riding a bicycle (Los Angeles to Canada to NYC to Key West...), carrying all my stuff (slept rough). I never changed or cleaned or lubricated anything. I assume the drive chain pieces all wore each other out equally and kept meshing, sort of, for thousands of miles, fully loaded, over 14K feet elevation, dirt roads... I wasn't picky about performance and didn't mind small noises. I wasn't stranded anywhere. I'm pretty sure I wasn't riding as quickly or efficiently as Lance. I didn't care.

But it seems that the right thing to do is replace chains before they stretch enough to wear out the teeth of chainrings and cogs; if chains were more expensive than chainrings and cogs I think it would be the other way around. Now I live in town, clean and lubricate my chain, replace it at 0.5% stretch. I sometimes I think I'd be happier if I rode until parts broke then replaced the whole drivechain.

You'll note that in Sheldon's/John's article and many others, the recommendation is actually to replace chains at 12-1/16" in order to prevent the next chain from skipping on the sprockets, and that if it gets to 12-1/8", you've probably trashed the cassette beyond usability with a new chain. If you search this site for "chain rotation", there are some members who feel that as long as you maintain a cleaning/relubing cycle with multiple chains, you can rack up a lot more total miles and go past 12-1/8" without encountering chain skip or tooth breakage.

(I'm currently trying out this method, but haven't ridden nearly enough miles -- 25,000 by some claims -- to prove that it works for myself, yet. )

ThermionicScott: I think chain rotation is a good idea. If I find a really good deal on a chain I know I like, I will stock up for sure. But for now, I've worn the chain to 12-5/32", so I'm going to try a new chain and see how I do. I keep a relatively high cadence, and, I've let the chain wax get squeeky a few times, so I think the chain wore faster than usual, so the wear ramps on the teeth are a little more steep than if I had a chain with a long life. I'll try to remember to report back when I change the chain.

4 months of use later, I changed the chain! It doesn't skip! I used a relatively cheap SunRace hollow pin chain. Chain is worn to a hair below 12-6/32"

There are two basic schools of thought. Then old school is to run a chain and sprockets until you can't. The prevalent approach these days is "preventive" replacement of chains before the sprocket is too far gone. You may believe that then wear pattern is such that a new chain will work even after the prior chain is "stretched" 1% (1/8"over 12"), but the cumulative experience of thousands upon thousands of riders indicates otherwise.

BTW- your theory that the stretched chain rides out toward the tips of the teeth is flawed. This applies to chains running at high speed and low torque, A bike chain is opposite, running at low speed with high load. The engaged rollers reach maximum load at about 11:00 as they're rising toward the top. At the moment the chain isn't pulling at a tangent, but on a secant, pulling the roller in and down to the root of the tooth. It then does roll up nearer to 12:00 as it approaches disengagement, but it's the root wear, not the tip wear that causes skipping.

Um... that would be, thousands of riders, minus one, if you want to be fully accurate.
Also, it wasn't my theory that the chain rides outwards on the teeth: I got it from Sheldon Brown's site.

My 8sp chain has about 600 miles of life left in it. It will have 17000 miles. The 3 smallest cogs will be replaced then.
I remove my chain every 800 miles and clean it in an ultrasonic cleaner with Simple Green and water. I lube my chain with chain saw bar oil. The chain will be at 1/16" wear.

Today I had an especially hard ride, and I noticed the chain skipped on occasion. I was so pleased the first day that I didn't notice any skipping. Oh well. That makes it 1001 people with skipping chains. I can live with it though. I'll see if the skipping goes away with time.
The link at 11 doesn't know about the rest of the chain. It is pulled by the link in front of it, and while that isn't tangent, it isn't at all the chain links going toward the chainring at the top.

I bet that's an impressive amount of miles; I don't measure mileage as I don't use a computer. Does the chain saw bar oil end up looking black on the chain after a couple weeks?
Instead of an ultrasonic cleaner, I use mineral spirits, or even paint thinner, and shake it in a bottle.

I rotate chains, 3-4 chains, 1 cassette, 0,5 middle ring. At 0,5% wear I thoroughly clean the chains with gasoline, dry them and heat them up in a vessel with chain saw oil. This besides the regular lubing and cleaning. Even then I get less than 17000 miles out of three-four chains. I might be doing something wrong. I do ride my bike in all weather conditions and a lot of road grime in the winter.

As with chains and lube: YMMV .

Yes, that's basically what I said, and I'm not sure if you misunderstood or not. The wear happens between 11 and 12 o'clock or so as the line of pull swings from a secant to the tangent. With a new chain and sprocket the wear ends there as the tooth drops clear. But with a worn chain, the wear continues as the tooth passes 12:00 and the roller rolls up the back while disengaging. Either way, wear happens during the moment of disengagement under load, with the wear rate increasing with sloppy (worn) fit.

I used to clean them in a Ragu bottle with mineral spirits, but the ultrasonic cleaner does so much better that my chain life has more than doubled.
The black will always come out of the chain. I wipe it off and it is less of a problem.

That wear will happen higher and higher up the tooth. The distance between the teeth stay the same no matter what the wear. As the chain wears, it gets longer, and as a consequence, ride higher on the teeth, because higher up on the teeth, the diameter of the cog is greater. I just found this video, and when tension is applied, all the chain links ride up the teeth, so that the distance between the links match the distance between the teeth (see starting 1:09): https://www.youtube.com/watch?v=a-7SWfQeqXI
Since the human power stroke is not even, there is probably wear happening at some range of diameter along the teeth, throughout the circumference. Add to that the secant to tangent wear you mentioned, between 11 to 12 o'clock.

edit: Oops, I was thinking of a worn chain. It is actually a new chain being put on an old cassette. I imagine the ramps on the cogs have been worn to slant a few degrees clockwise towards tangent. Instead of the secant force being perpendicular to the surface of the tooth, it is now at an acute angle, and the chain wants to ride up under force; the chain elongates under this force, to a higher diameter.

On a tangent thought (pun), imagine the shark fin look of a worn tooth.