The results of shoulder arthroplasty for osteoarthritis have been reported to be excellent or good for the majority of patients, but the value of using a glenoid component and the anatomic factors that affect outcome are still debated. The purpose of this study was to evaluate the influence of an operatively confirmed full-thickness tear of the rotator cuff, the severity of preoperative erosion of glenoid bone, preoperative radiographic evidence of subluxation of the humeral head, and the severity of preoperative loss of the passive range of motion on the outcome of total shoulder arthroplasty and hemiarthroplasty. In a multicenter clinical outcome study, we evaluated 128 shoulders in 118 patients with primary osteoarthritis who had been followed for a mean of forty-six months (range, twenty-four to eighty-seven months). Patients with <10 degrees of passive external rotation preoperatively had significantly less improvement in external rotation after hemiarthroplasty (p = 0.006). Thirteen (10%) of the 128 shoulders had a repairable full-thickness tear of the supraspinatus tendon, but these tears did not affect the overall American Shoulder and Elbow Surgeons score, the decrease in pain, or patient satisfaction. Severe or moderate eccentric glenoid erosion was seen in twenty-nine (23%) of the 128 shoulders, and total shoulder arthroplasty resulted in significantly better passive total elevation and active external rotation as well as a trend toward significantly better active forward flexion than did hemiarthroplasty in these shoulders. The humeral head was subluxated posteriorly in twenty-three shoulders (18%), and when they were compared with the other shoulders in the study, these shoulders were found to have lower final American Shoulder and Elbow Surgeons scores, more pain, and decreased active external rotation following either total shoulder arthroplasty or hemiarthroplasty. On the basis of our data, we recommend the use of a glenoid component in shoulders with glenoid erosion. Humeral head subluxation was associated with a less favorable result regardless of the type of shoulder arthroplasty and must be considered in preoperative planning and counseling. Severe loss of the passive range of motion preoperatively was associated with a decreased passive range of motion postoperatively. A repairable tear of the supraspinatus tendon is not a contraindication to the use of a glenoid component. .