For those of you who don't not understand that metals deform by dislocation glide, please skip to the last paragraph

The fun bit about fatigue is that is is intrinsically linked to crystallographic orientation. Perfect metals - those with a face-centred-cubic crystal structure and the maximum possible slip modes do not have a fatigue life due to the ease with whihc you can deform them: It doen't matter what defomation vector or combination of vectors are applied, they will continue to platically deform until the materil runs out of dislocations.

Hexagonal metals, and body-centred cubic metals have a defined fatigue life because below certain dislocation glide stresses, the dislocations present within the metal will not budge. Either they are locked by having a limited number of directions they can move through the metallic lattice or insufficient stress to activate them has been applied.

Hexagonal metals have limited dislocation directions, but low activation stresses, body-centred-cubic metals have high activation stresses below certain temperatures, and limited dislocation directions. As an aside, this is what makes ferritic steels brittle at cryogenic temperatures.

Skip to: Magnesium also has a fatigue life, no matter what that article says. It's hexagonal, with a high c/a ratio. It's just a very low stress limit fatigue life.