The 1992 NIH Research Planning Conference on Prosthetic and Orthotic Research for the 21st Century (Childress, 1992) recognized that the field of prosthetics lacks theoretical understanding and empirical studies on learning to control an upper-extremity prosthesis. We have addressed this problem using a novel approach in which persons without amputations are required to perform or learn basic motor tasks using a prosthetic simulator. The findings so far have three important implications for rehabilitation: (a) learning prosthetic control can be facilitated by general principles that have been shown to promote motor learning; (b) bilateral transfer can be used to enhance prosthetic limb control and minimize rejection; and (c) a voluntary closing terminal device has some advantages relative to a voluntary opening device. (Contains 1 figure.)