A large driving simulator experiment is carried out, comprising a medical/neurological and neuropsychological evaluation of the participants, and a set of driving tasks for different scenarios. The participants comprise two distinct groups from the general population: a group of individuals with mild pathological conditions, and a group of healthy individuals. Models of driver distraction are then developed in order to identify and rank the causes of distraction. Models of driver behaviour and safety are also developed, in order to quantify the impacts of distraction. Driver speed and vehicle lateral position are modelled in relation to distraction, road / traffic conditions, driver characteristics etc. Moreover, driver reaction time and accident probability are modelled within the occurrence of unexpected incidents. Measures for the improvement of driver behaviour are finally proposed.