Fifty-six Dana unconstrained total shoulder arthroplasties in forty-seven patients were followed for a minimum of two years. The preoperative diagnoses included osteoarthritis, rheumatoid arthritis, traumatic arthritis, avascular necrosis, failed hemiarthroplasty, and failed total shoulder arthroplasty of another design. For all of the patients, the average rating for pain improved from 3 points preoperatively to 8 points postoperatively, and the average rating for function improved from 3 to 7 points. The ranges of abduction and of external rotation improved substantially in the shoulders that were affected by osteoarthritis, rheumatoid arthritis, or osteonecrosis. Ten patients were treated with a hooded glenoid component, designed to improve stability in shoulders in which the rotator cuff is deficient. In these shoulders, both the rating for pain and the rating for function improved, although the range of motion did not. Complications that required revision of the arthroplasty developed in five shoulders in which a regular component had been implanted and in two that had a hooded glenoid component. A radiolucent line developed at the bone-cement interface of the glenoid component in fifty-three shoulders, but only two revisions were done for loosening of the glenoid component.