The development and validation of Advanced Driver Assistance Systems (ADAS) and automated driving functions constitutes a challenge, regarding the prospective evaluation of the system’s impact on traffic and traffic safety. It requires an assessment methodology, which can perform the quantification of both benefits of risks in a prospective way. An approach of simulation-based virtual experiments is able to fulfil these requirements. To this end, key traffic processes are modeled with respect to microscopic interactions between traffic participants in a large variety of possible traffic scenarios including relevant boundary conditions (e.g., environment). Traffic processes are represented by detailed, stochastic models of drivers, vehicles, and road environment, together with their interactions. In the present paper, a numerical driver behaviour model is introduced, which is applicable for the prospective safety impact assessment and has been cooperatively developed by TU Dresden and BMW. It is capable to represent the vehicle following behaviour of an unassisted driver, covering normal driving situation, critical driving situation, up to rear-end collision. The driver behaviour is sequentially simulated within three submodels, which can also influence each other by feedback mechanisms. The first submodel provides the acquisition of information and considers the selective attention of the driver by means of a stochastic field of vision motion algorithm and a mental environment model. Only information perceived by this first submodel can be processed and assessed by the second submodel. The thereby derived decisions are determined by a reaction time model and are finally translated into discrete action patterns by the third submodel. To consider both intra- and inter-individual variability of driver behaviour, the internal processes and accordingly the behavioural outputs are influenced by stochastically parameters. By using the driver behaviour model, the unassisted surrounding traffic can be simulated in a realistic way. Therefore, the impact of ADAS and automated driving functions can be assessed in microscopic interaction processes in a dynamic traffic context. Furthermore, the simulation results can also be used to optimise the parameters of the optimised function to maximize the safety benefits. In the near future, applying such driver behaviour models to predict the behaviour of the surrounding traffic is also conceivable.