Analysis of the circumstances surrounding motorway accidents clearly points to drowsiness at the steering wheel as one of the main causes. It is therefore not hard to understand the growing interest shown by the car makers in rapidly developing a system for detecting lapses of alertness. Renault's Département Biomédical de l’Automobile recently organized a series of tests to validate such a system. 21 tests were performed on a driving track at night. Numerous lapses of alertness were obtained, leading to drivers often dozing off and leaving the road. This paper presents the results of the physiological recordings (electroencephalogram, electrooculogram) and the behavioural recordings (analysis of the video film of the driver's face). It is shown how we establish a physiological reference of the driver’s level of alertness from this data. The correlation between the results provided by the EEG and those obtained by behavioural analysis is discussed. We look at our multisensor approach, based on the analysis of driving behaviour, and the respiratory signal as well as the automatic processing of the film of the driver's face. The study of the driver's reactions based on mechanical signals collected on the vehicle is also briefly described.