Training strategies to reduce the well-documented (e.g., Lim & Petersen, 1985; Voyer, Voyer, & Bryden, 1995) gender difference in visual spatial ability were investigated. Participants (21 men, 21 women) were trained under 1 of 2 conditions to estimate the orientation angle of a ship viewed on a submarine periscope simulator. The data were analyzed in a 3 by 2 by 2 (Training Conditions by Gender by Test Session) mixed design with repeated measures over the last variable. A significant Training Condition by Gender interaction revealed large performance differences in favor of men in the control group and the group trained using an instruction manual. However, no significant gender difference was found for participants trained with repeated practice accompanied by feedback. This suggests that even a brief training session, using an appropriate instructional strategy, may be all that is required to increase the mental rotation performance of women to the level of men on a specific task. The benefit of the instruction was still evident when tested 3 weeks later; the gender difference was still absent.