The development of alerting driver assistance systems and the evaluation of their effectiveness constitute a great challenge, because this effectiveness depends on the driver’s reaction. Thus, a new evaluation method based on the numerical simulation of that driver reaction is developed at the Chairs of Traffic and Transportation Psychology and Automobile Engineering of the Technische Universität Dresden. The driver model is able to simulate the event chain of hazard cognition (information acquisition, information analysis and decision) and driver behavior (action selection and implementation) by means of traffic-psychological submodels, both while driving uninfluenced and when supported by the driver assistant systems. The underlying reaction time model and physiological mechanisms result in varying parts of the driver reaction both statistically and dynamically within each trial. Hence, the whole procedure can be described as a virtual study with human participants. Based on the overall results of these simulations, the mitigation of accident severity or respectively the chance in collision avoidance resulting from the application of the evaluated safety concept will be able to be assessed in the near future. Thus, this method will be equally suited as a tool for the development and for the evaluation of such systems. The subject of the present paper is the introduction of the driver model by means of its logical and structural layout. The concept of the signal processing chain and the relation to the underlying reaction time model is to be explained by describing a direct parameterization process from simulator and test track studies with human subjects. A full concept of evaluation metrics for systems of active safety by using the driver model will additionally be introduced. The necessary steps of that implementation will also be discussed.