In this study, the accident simulation tests, the so-called crash tests, enforced by legislation and put into praxis are evaluated in regard of their conformity with reality. They are based on accident analyses from investigations at the place of accident, which are carried out by a scientifically trained team which documentates details of the accident event. 826 cars involved in traffic accidents with cars, trucks and other objects, in the greater vicinity of Hannover (FRG) were at our disposal for evaluation purposes. The study clearly reveals that impact simulations like those carried out at present, cover only approximately 34% of all situations of road traffic. This conclusion is derived from a comparison of accident framework conditions like overlapping degree, impact impulse direction and impact situation. For the frontal impact the offset impact, with two-third degree of overlapping, without rail-bound lead should be favoured. The lateral collision should be simulated with oblique and eccentric impacting barriers from the front.