Compared to the Passive Safety of road vehicles, Active Safety is a rather young discipline in automotive engineering, as a remarkable traffic penetration was just established in the 1990s by a broader application of reactively intervening suspension control systems like ESP and ABS. Advanced driver assistance systems (ADAS) which should avoid critical situations before they emerge by proactively interventions are serially produced for about ten years now. The evaluation of these ADAS’ impact on the future accident statistics constitutes a huge challenge yet, because the vast amount of possible interactions with the other elements of the traffic system is hardly quantifiable in a prospective manner. Therefore, mostly aspects of functional safety are considered today in the development and assurance of such systems. An integral evaluation approach becomes inevitable, to take the whole potential of these ADAS into account within the assessment of consumer protectors, or to adjust future concepts most precisely on real traffic conditions. Such an integral evaluation approach is outlined in the present work, which especially uses numerical methods in the early development phase. The behaviour simulation of traffic participants is identified as the largest gap in this field. The aim of the present work is therefore the development of a first driver behaviour model, which is suitable for the prospective, stochastic evaluation of an ADAS’ impact on traffic safety. This is prototypically realized within a pedestrian crossing scenario and a generic collision avoidance system. Also, the virtual driver is only able to brake in this first application. At first, an extensive overview on literature is given in chapter 2, which is needed for the modelling task described above. Besides, the existing gaps in literature are identified. Those gaps are subsequently addressed in chapter 3 by human subject studies, so the driver behaviour model can be constructed in chapter 4. Finally, the stochastic behaviour of the model is reviewed in chapter 5 by comparing its simulated behaviour to another human subject study. Due to methodical issues of this study, the correctness of the model behaviour cannot be explicitly confirmed. Nevertheless, a prototypical evaluation method of the generic ADAS is presented in chapter 6 by introducing the Virtual human subject study. For one sample with ADAS and one sample without ADAS of 66,564 subjects each, simulations are executed. By means of statistical tests, significant differences in reaction time, collision frequency, and velocity reduction due to the application of the ADAS are proved. For example, the collision frequency can be reduced by 51.9 %.