A new method of presenting simulated wheel/rail forces and relating these to the observed wheel damage has been developed. This indicates a good correlation between the predicted forces and the regions of the wheel where damage is observed in practice. There is also a good correlation between the angle of the predicted forces and the observed cracks. The angle evidence suggests that the dominant rolling contact fatigue cracks on the field side of the wheel tread are initiated by the occasional high forces when the opposite wheel is running in flange contact on sharp curves. Cracks may then be propagated by more frequent lower forces on moderate curves.