Two novel cadaveric models of a thrower's shoulder were evaluated. Both models included the capsuloligamentous and osseous components of the glenohumeral joint. In model 1 the coracohumeral ligament was retained and the glenoid was positioned parallel to the ground whereas the humerus was positioned at 60 degrees of glenohumeral elevation in the scapular plane. In model 2 the coracohumeral ligament was resected and the glenoid was positioned vertically at 30 degrees of scapular elevation whereas the humerus was positioned at 60 degrees of glenohumeral elevation in the scapular plane. Each specimen was sequentially tested under 3 conditions: intact, after nondestructive anterior capsular stretch, and after simulated posterior capsular contracture. Measurements included rotational range of motion and relative glenohumeral position from neutral to maximum external rotation of the humerus. In model 2 the glenohumeral forces were also measured by use of a 6-degree-of-freedom load cell. The stretching of the anterior capsule in model 1 was performed while the humeral head was constrained in the glenoid and in model 2 while the humeral head was unconstrained. Both models showed increased humeral external rotation and decreased humeral internal rotation similar to that seen in throwing athletes. Both models also showed an increased humeral shift inferiorly after anterior capsular stretching. Model 1 demonstrated a subsequent humeral shift superiorly after posterior capsule plication. Unconstrained specimens of model 2 dislocated posteriorly and inferiorly after posterior capsule plication when the humerus was rotated to maximum external rotation. On the basis of these findings, for future studies, constrained capsular stretching models will be further investigated.