A prospective study of displaced femoral neck fractures was conducted with patients randomized into three groups: cemented total hip replacement (THR), cemented hemiarthroplasty (UHR-C), and noncemented hemiarthroplasty (UHR-NC). Eighty-nine patients were treated for displaced (Garden type 3 or 4) fracture with a prosthesis; 39 patients had THR, 37 had UHR-C, and 13 had UHR-NC. No difference was found in the level of pain, ambulation, or aids required between the THR group and the UHR-C group, except for active community ambulators, who demonstrated decreased endurance capability when treated with UHR-C. The patients treated with UHR-NC had increased pain and decreased ambulation and required assistive devices. The most significant complication was dislocation, which occurred in 7 patients with THR and 2 with UHR-C. The mortality rate at 2 years was 18%; all but two of the deaths were in patients with one of five different medical diseases: clinically significant cardiac or pulmonary disease, cancer, alcoholism, or psychosis. The mortality rate was insignificant in patients younger than 70 years of age. Roentgenographic review showed only 2 of 76 cemented stems with progressive cement-bone demarcation, with one patient requiring revision for a loose prosthesis. No evidence of protrusio was seen with the use of the bipolar cap. The uncemented hemiarthroplasties demonstrated lucency between the metal and bone in the proximal femur in 9 of 13 stems. The authors make recommendations regarding hip replacement.