The aim of this study is to compare semi-automatic and hand driving efficacy and to identify psychophysiological indicators of potentially dangerous driver’s states which can be used before the trip or in real time. 32 subjects (18–22 years old) participated in the study after partial sleep deprivation (an average of 5.8 hours of sleep the night before the experiment). Each subject was tested in two 1.5-hours long experimental sessions in driving simulator with hand and semi-automated computer driving simulator with three event types. They were constantly required to respond to two types of stimuli: gradually changing, appearing unexpectedly, and requiring maneuvering. Simulation data and EEG were recorded using a standard 10–20 system with a sampling frequency of 500 Hz. A 3D video camera was fixed on the monitor in the subject’s room, which tracks the position of the head and changes in the subject’s facial expressions throughout the experiment.The subject took questionnaires for drowsiness and general well-being, and the reaction time was also tested before the experiment. With manual and semi-automatic control, 1.2 and 4.3% of all stimuli were missed, respectively. Mean reaction time for slowly changing stimulus was 4.29 ± 0.09 и 3.19 ± 0.12 seconds, respectively. Reaction time was not changed significantly for sudden stimulus, but much more lapses were found in semi-automated sessions. For semi-automated sessions, 21% of lapses were preceded by normal blinks, while for 75% at least one prolonged blink (>0.9 sec) was found within preceded 10 sec interval. A significant correlation (40–50%, p < 0.01) with the average reaction times in each session was shown by the results of the simple reaction time test performed before the experiment. These results give us further evidence that both drowsiness and distraction contribute to efficiency loss during semi-automated driving and that subject’s state should be controlled continuously during driving.