Abstract
A rood traffic safety policy must have the results of a methodical well done scientifically research. The traffic safety problem must be multidisciplined, whereby a more dimensional probability model can give the best results of the research. Measures to improve the traffic safety with the aid of the three E's (Engineering, Education, Enforcement) must be based on objective facts. To describe and to research the problem of traffic safety it is needed to adapt the registration of traffic accidents to the needs concerning the objectivities and the reliability of the registration.