Eighty college students were subjected to 0, 2, 5, or 10 simulated accidents on a modified bench grinder to test the long-term effectiveness of accident simulation as a technique for teaching safe operating procedures on small power tools. Half the subjects in each group received the simulated accidents on consecutive trials during the 15 training session trials, while the other half received the same number of simulated accidents intermittently throughout the training session. Retention of the safety training was tested after 1, 3, and 6 months. The results showed that 6 months after the single training session all groups trained with the use of simulated accidents had significantly fewer "accidents" than the control groups given traditional safety training. Neither the number of simulated accidents nor the method of presentation made a significant difference in the number of accidents, at least not over the 6 month duration of this study. It is concluded that accident simulation is a promising technique for teaching safe operating procedures on small power tools.