The present study investigated the effects of discomfort glare on driving behavior. Participants (old and young; U. S. and Europeans) were exposed to a simulated low-beam light source mounted on the hood of an instrumented vehicle. Participants drove at night in actual traffic along a track consisting of urban, rural, and highway stretches. The results show that the relatively low glare source caused a significant drop in detecting simulated pedestrians along the roadside and made participants drive significantly slower on dark and winding roads. Older participants showed the largest drop in pedestrian detection performance and reduced their driving speed the most. The results indicate that the deBoer rating scale, the most commonly used rating scale for discomfort glare, is practically useless as a predictor of driving performance. Furthermore, the maximum U. S. headlamp intensity (1380 cd per headlamp) appears to be an acceptable upper limit. © 2002, Human Factors and Ergonomics Society.