Because bone is obviously in some way adapted to the loads falling on it and because fracture is usually the failure of mechanical competence of main clinical importance, it is often thought that bones are adapted to resist fracture. In this perspective, I consider that this may not be the case. Bones may be designed to be very stiff, and therefore highly mineralized, and therefore brittle; they may be adapted to normal loads, but not to the characteristic loads occurring in falls, or may be very poorly designed to stop cracks traveling once they have started. Bones may also potentially fail in completely contrasting modes, and therefore their design has to be a compromise that does not resist either mode completely successfully. The greatly differing fracture incidences in different bones seen in pre-senile adults suggest that safety factors have been adapted, over evolutionary time, to produce the best compromise for a host of different design constraints.