Seventeen long haul truck drivers were recruited to participate in a driver fatigue study. The study was conducted to determine if a degradation in driver performance could be detected from controlled inputs of drowsy drivers. A test protocol was developed to replicate the kind of driving environment generally regarded (and often associated) with single vehicle run-off-the-road type crashes. Physiological measures of drowsiness were recorded while a fully instrumented 3-axle truck-tractor was driven on a closed-circuit track. Driving sessions were conducted when test subjects were both alert and sleep deprived. Analysis and refinement of experimental data were used to develop a preliminary steering-based algorithm. The results of the study indicate that lane departure, arising from a loss of alertness due to fatigue, may be predicted by monitoring movements of the steering wheel.