Background. While the number of fatal accidents decreases every year, there is still need for improvement and action to prevent these deaths. The basis for this purpose has to be an analysis about the factors influencing the car crash mortality. Patients and methods. This multivariate analysis is based on 2,609 cases collected by the accident research units in Dresden and Hanover during the years 1999-2003. This paper presents a multivariate model containing ten different variables which detects 93% of these cases properly as "survived" or "dead." Results. The variable "energy equivalent speed" (EES) has the most important effect on car crash mortality. Values greater than 60 km/h lead to a 248-fold higher mortality risk. The mortality is not related to gender, but rather to the value of EES. Younger persons (18-25 years) have no significantly higher risk for death, but for persons older than 60 years the risk was threefold higher. Conclusion. Univariate assessments may be falsified by bias; this effect can be uncovered by multivariate models.