When using eye movements to determine the state of a car driver it is important that the eye tracker is robust, unobtrusive, inexpensive and fully automatic. The objectives of this study are to compare the performance of a one-camera system with a three-camera system and to investigate if the accuracy and availability of the one-camera system is sufficient to monitor driver state. Data from 53 subjects were evaluated and the results indicate that there is not much difference between a single-camera system and a multi-camera system as long as the driver is looking straight ahead. However, with more peripheral gaze directions, the larger coverage that is provided by the additional cameras works in favour of the multi-camera system.