This paper is concerned with computational driver modeling, whereby a particular focus is placed on mapping both non-routine and routine elements of the driving task in a theoretically coherent framework. The approach is based on Salvucci’s [1] driver model and thus, the cognitive architecture ACT-R [2] is used for modeling non-routine matters; for routine activities, such as the longitudinal and the lateral control of the vehicle, a fuzzy logic approach is suggested. In order to demonstrate the applicability of this procedure, an empirical evaluation study is carried out and the steering behavior of a computational driver model is compared to that of human drivers.