Seventy-eight Neer hemiarthroplasties and thirty-six Neer total shoulder arthroplasties were performed in patients aged 50 years or younger between January 1, 1976, and December 31, 1985. Sixty-two hemiarthroplasties and twenty-nine total shoulder arthroplasties with complete preoperative evaluation, operative records, and a minimum 15-year follow-up (mean, 16.8 years) or follow-up until revision were included in the clinical analysis. Sixteen patients died, and seven were lost to follow-up. All 114 shoulders were included in the survival analysis. There was significant long-term pain relief (P < .01) and improvement in active abduction (P < .01) and external rotation (P < .01) with both procedures. There was not a significant difference between total shoulder arthroplasty and hemiarthroplasty with regard to pain relief, abduction, or external rotation. Radiographs were available for 53 hemiarthroplasties and 25 total shoulder arthroplasties with a minimum 10-year follow-up. Humeral periprosthetic lucency was present more frequently after total shoulder arthroplasty (60%) compared with hemiarthroplasty (34%) (P = .0079). Glenoid erosion was present in 38 of 53 hemiarthroplasties (72%). Glenoid periprosthetic lucency was present in 19 of 25 total shoulder arthroplasties (76%). The results were graded by use of a modified Neer result rating system. Among the hemiarthroplasties, there were 6 excellent (10%), 19 satisfactory (30%), and 37 unsatisfactory results (60%). Among total shoulder arthroplasties, there were 6 excellent (21%), 9 satisfactory (31%), and 14 unsatisfactory results (48%). The estimated survival rate for hemiarthroplasty was 82% (95% CI, 74%-92%) at 10 years and 75% (95% CI, 64%-86%) at 20 years. The estimated survival rate for total shoulder arthroplasty was 97% (95% CI, 91%-100%) at 10 years and 84% (95% CI, 68%-98%) at 20 years. The data from this study indicate that there is marked long-term pain relief and improvement in motion with shoulder arthroplasty. However, there is a moderate rate of hemiarthroplasty revision for painful glenoid arthritis. Unsatisfactory result ratings were most commonly a result of motion restriction from soft-tissue abnormalities. Great care must be exercised, and alternative methods of treatment considered, before either hemiarthroplasty or total shoulder arthroplasty is offered to patients aged 50 years or younger.