We evaluated the effect of two different anterior capsular shift strategies on load-induced multidirectional glenohumeral motion. Nine cadaveric shoulders were tested on a special test apparatus in which three motions were constrained but which allowed simultaneous measurement of three translations: anteroposterior, mediolateral, and superoinferior. A simulated Bankart lesion was created by detaching the inferior glenohumeral ligament and labrum from the anterior glenoid from 5 mm superior to the anterior band of the inferior glenohumeral ligament to a point just posterior to the infraglenoid tubercle. The two types of medial-based anterior capsular shifts were performed sequentially using fixation via a transglenoid drill hole and No. 2 suture. One shift medialized the anterior capsule on the glenoid by 5 mm; the other one shifted the anterior capsule 5 mm superior on the anterior glenoid. Biomechanical testing was done in two positions of humeral elevation in the scapular plane, three positions of humeral rotation, and with an externally applied joint compression load of 22 N. Glenohumeral motion was measured in the intact state, after creation of the simulated Bankart lesion, and after each of the two anterior capsular shifts. Both the superior and medial shifts of the capsule decreased anterior glenohumeral translation to equivalent extents. For posterior and inferior translation at 45 degrees elevation, the superior shift significantly decreased translation to a greater extent than did the medial shift.