Principles of an infrared night vision system in cars were studied in a stationary car simulator. The 20 participants drove for about 45 min on a rural route in night conditions where four objects were placed in or next to the road. When the camera was mounted on the grill of the car and the scale was true size, that is, 1:1, the time to collision was the longest. Changing the scale to a wide-angle view of 1:2 decreased this time. When the camera was mounted on the roof of the car, the scale change produced an increase of the time to collision. Principal omponents analysis indicated that the speed and the lateral position of the car were independent. The results were explained partly in terms of differences between the optic arrays presented to the drivers and to motion adaptation. It is suggested that the camera for a night vision system should be placed at the level of the car grill and the presentation scale should be 1:1.