We conducted this cadaveric study to define a biomechanical rationale for rotator cuff function in several deficiency states. A dynamic shoulder testing apparatus was used to examine change in middle deltoid muscle force and humeral translation associated with simulated rotator cuff tendon paralyses and various sizes of rotator cuff tears. Supraspinatus paralysis resulted in a significant increase (101%) in the middle deltoid force required to initiate abduction. This increase diminished to only 12% for full glenohumeral abduction. The glenohumeral joint maintained ball-and-socket kinematics during glenohumeral abduction in the scapular plane with an intact rotator cuff. No significant alterations in humeral translation occurred with a simulated supraspinatus paralysis, nor with 1-, 3-, and 5-cm rotator cuff tears, provided the infraspinatus tendon was functional. Global tears resulted in an inability to elevate beyond 25 degrees of glenohumeral abduction despite a threefold increase in middle deltoid force. These results validated the importance of the supraspinatus tendon during the initiation of abduction. Glenohumeral joint motion was not affected when the "transverse force couple" (subscapularis, infraspinatus, and teres minor tendons) remained intact. Significant changes in glenohumeral joint motion occurred only if paralysis or anatomic deficiency violated this force couple. Finally, this model confirmed that rotator cuff disease treatment must address function in addition to anatomy.