The effectiveness of an industrial lift truck operator safety training program utilizing behavior sampling procedures was investigated. The basic training approach specified needs assessment, program development, and program evaluation. Needs assessment involved the determination of existing knowledge concerning lift truck safety and injury data analysis. Program development involved a detailed task/hazard analysis to derive a rational basis for the training program. Program evaluation was based on results of on the job behaviors observed before and after the training. Two studies, each employing a different experimental design, were performed at large distribution warehouses. Study 1 utilized a between and within groups comparison design. Operators were stratified by vehicle type, department, and relative exposure. All lift truck operators were randomly assigned to training only, training plus feedback, and control, which received no training or feedback. Study 2 was modified experimental design employed to eliminate any partially mitigating influence of experimental groups interacting in conflict avoidance situations. Following initial training, all three groups in Study 1 showed a decrease in mean error rates with the training plus feedback group showing the largest decrease (23 percent), followed by training only (18 percent), and control (6 percent) groups. Total performance gain demonstrated in Study 1 was 44 percent improvement from pretraining measurements. Twelve of 14 individual behaviors used for performance evaluation showed improvements. The overall net improvement in mean error rates due to the training and feedback program was stronger in Study 2 (70 versus 44 percent). Similar to Study 1, no single behavior dominated the observed effects. The authors conclude that a well designed and administered occupational safety training program, emphasizing safe work practices and derived from a true assessment of need, can be effective in improving on the job behavior.