Conservative management of rotator cuff pathology often involves certain therapeutic exercises. Although a major goal of these exercises is to increase strength of the rotator cuff, little empirical evidence supports this assertion. In this study, 34 nonpathologic young adults were pretested using a LIDO Multijoint II isokinetic device for average and peak torque generated during internal and external rotation. Subjects were arbitrarily assigned to a right-arm- or left-arm-trained group, exercised for 4 weeks, and then posttested for changes in humeral rotation torque. Moderate but significant increases in torque (8-10%) as well as in total work done were observed in both groups, only in the trained arm. Subjects who trained the nondominant (left) arm experienced gains similar to those who trained the right arm. Gains were significant in the case of both internal and external rotation (also average as well as peak torque), with men and women experiencing the same relative increases. These data, in addition to supporting the use of selected exercises to increase humeral rotation torque in a healthy population, offer a potential model for the rehabilitation of patients with rotator cuff injury.