Originally Posted By: KrisZ
As mentioned before, steel bolts don't fatigue, unless they are torque to yield. Torque to yield takes the material form its elastic range (that's where it always comes back to the original length), to the plastic range (that's where the material stretches beyond its ability to spring back).
Once the bolt goes into the plastic range, the structural changes are irreversible and it is only a matter of time when the bolt fails.
That is a very good explanation. If steel is kept in its elastic range it does not fatigue.
There are a couple of other possible contributors. The first is stress corrosion cracking where a corrosive agent (salt water for one) in combination with a repeated stress may lead to a crack which over time grows and extends across the bolt, until it suddenly fails under an otherwise normal load.
The second is that a stress riser, perhaps at a thread or a deep scratch, perhaps at a large flaw in the metal, results in local increased stress sufficient to start a crack which progresses in much the same way.
An expert can tell a lot by looking at the broken end. If it looks like a piece of toffee torn apart it was a plastic failure, if there is a coarse surface it was a brittle failure, if a large flat area with concentric "ripples" a fatigue failure.