The purpose of this study was to analyze pre-operative and intra-operative factors that affect the outcome of shoulder arthroplasty. We undertook a retrospective review of all shoulder arthroplasties performed at our institution between 1986 and 2003. Patients were contacted and outcomes were assessed using the Simple Shoulder Test and the Western Ontario Osteoarthritis of the Shoulder Index questionnaires. One hundred six patients (126 shoulders) participated in the study. The average length of follow-up was 6 years 9 months (range 2 to 20 years). Revision arthroplasty surgery and female gender were associated with worse outcomes. Age, the number of medical comorbidities, obesity, pre-operative range of motion, prior non-arthroplasty surgery, smoking, and alcohol abuse did not correlate with outcome. Patients who had shoulder arthroplasty for osteoarthritis had better outcome scores than those with rheumatoid arthritis. For intra-operative variables, significantly worse outcomes were found both with the use of hemiarthroplasty and in patients with a rotator cuff tear identified at the time of surgery. These findings may help to optimize patient and surgery selection in shoulder arthroplasty and assist in preoperative patient counseling.