There is a view that internal corrosion can eventually make them fail to operate as they are supposed to, in an overcurrent condition.
This is why I don’t accept old sold as new CBs.
However one cannot tell at what age the problems might start. Is there any data out there?
I think reputable avionics shops replace CBs routinely when doing large upgrades, but usually not all of them because it can be a multi-day job to change them all.
I believe that actually using the CB, that is having current going through, etc., probably does “wear” them out over time.
I doubt that a CB sitting in a dry plastic bag on a shelf gets any wear.
I have also noticed that the higher the current, the faster they tend to fail.
Not sure if exercising them by pulling & resetting has any affect.
In some cases c.b. are used as on-off switches, either per SOP or e.g. for turning off turn gyro and avionic fan when powering up on the ground. Also when instructing, I make a point of pulling a c.b. sometimes, partly to simulate failures, and partly to show the student what a pulled/tripped c.b. looks (or feels) like.
I have indeed wondered if wear from regular use is something to consider.
Yes, they will fail faster. Is not hard to figure out why. There mechanical with parts that create sparks in metal as well if there under load.
It is not a good practice to turn on off your house electronics using the electrical fuse box.
I have also noticed that the higher the current, the faster they tend to fail.
That one I can see: thermal breakers work by getting hot and bending a bimetallic strip. If you have say a 10A CB, the chances are it is carrying say 5A and that will heat it up to a good portion of its tripping temperature. But if you have say a 1A CB, the chances are that it is feeding something like the stall warner which draws minus nothing, and 1A was used because it is the smallest CB rating. There are smaller ones but they cost more money 
and a CB is really mostly for wire (fire) protection, not for protecting the device(s) being fed.
Also when instructing, I make a point of pulling a c.b. sometimes
I wasn’t suggesting pulling them while loaded. I meant pulling them with the master OFF. Sorry I didn’t make that clear.
Pulling them while loaded will wear the contacts out. It might also have some other effects if they are already hot.
They must wear out eventually from being operated; I don’t know the rated life but it is probably in the hundreds. I would expect the domestic ones (which are magnetic) to be in the thousands. So exercising the breaker (when cold) say once a year should not shorten its life. But if you can’t pull it out with your fingers, it is probably seized up and won’t trip anyway.
I doubt that a CB sitting in a dry plastic bag on a shelf gets any wear.
Corrosion?
There is no such thing as a dry plastic bag, unless it is sealed and with silica gel inside. The 20 year old CBs which are being widely sold with all the fresh paperwork have been kicking around a variety of old aviation company / long defunct airline stores. The reason they surface now is because the last outfit which stored them went bust / didn’t want them anymore / thought they were junk 
And same goes for CBs installed in aircraft. The RH they see is the same as the outside RH, more or less.
Without going to the manufacturer, I don’ know if you can answer this question.
This HSE publication (pg 19, see excerpt below) suggests that for some relays, switching under the correct design load allows the resultant arcing to burn off oxide deposits. Conversely, with insufficient load, the oxide deposits build up and the relay could eventually remain ‘open’. This suggests that exercising a CB once in a while might not be bad.
From the UK HSE:
On the flip side, I have seen Klixon thermal reset breaker data sheets which suggest 500 manual operations or 2000 cycles. I am aware that some built in bimetallic strip devices might go for 10k to 20k cyles. If your CB is only good for a couple hundred cycles, I wouldn’t want to unnecessarily use it… I guess pulling it a couple of times a year is no big deal, but perhaps not everyday.
I have also seen heavily melted contact pads, presumably from switching under full rated (or higher) loads. But without doing detailed analysis of each case it is hard to know. Perhaps the contacts which failed were made of mystery metal, or perhaps corrosion had started and the contacts were high resistance, contributing to an accelerated rate of decay.
There is a bit of science about this, but I think that it is hard to know if we are using the lessons we learned 50-60 years ago, or if we have ‘forgotten’ them when we outsource the manufacturing.
Yes that is very true about relay contacts. One uses specific metals for contacts carrying very low currents (silver?). However I am not sure this is applicable to CBs, which spend their whole life in the normally-closed position.
One big difference between a thermal CB and a relay is that a CB that is carrying any significant current is constantly getting hot, and I think this is why they wear out. The materials don’t like the constant thermal cycling. Plastics especially don’t like it, and most of a CB has to be plastic. Whereas magnetic CBs can run without any heating, obviously.
Thermal CBs are a really crap idea, but they are cheap and compact and OK for wire (fire) protection.
