This study was conducted to determine the effect of in vitro passive and active loading on humeral head translation during glenohumeral abduction. A shoulder simulator produced unconstrained active abduction of the humerus in 8 specimens. Loading of the supraspinatus, subscapularis, infraspinatus/teres minor, and anterior, middle, and posterior deltoid muscles was simulated by use of 4 different sets of loading ratios. Significantly greater translations of the humeral head occurred both in 3 dimensions (P < .001) and in the sagittal plane (P < .005) during passive motion when compared with active motion from 30 degrees to 70 degrees of abduction. In the sagittal plane, passive abduction experienced a resultant translation of 3.8 +/- 1.0 mm whereas the active loading ratios averaged 2.3 +/- 1.0 mm. There were no significant differences in the translations that were produced by the 4 sets of muscle-loading ratios used to achieve active motions. This study emphasizes the importance of the musculature in maintaining normal ball-and-socket kinematics of the shoulder.