This paper presents the work that has been carried out within the SENSATION Integrated Project [1] concerning the prediction of sleep onset for drivers aiming to provide an effective measure for sleep related accident prevention. Indicators that have been proven to be the most relevant to the physiological manifestation of hypovigilance were examined and exploited in order to develop a holistic physiological sleep predictor that fuses eyelid activity and EEG features. The proposed algorithms were tested using driving simulator data from 35 subjects and the accuracy of the system was measured.