During the 11-year-period from 1970 to 1980, 799 total hip arthroplasties with metal-on-plastic prostheses were performed for primary osteoarthritis. In all cases high viscosity cement was used, but not under pressure; acetabular cartilage was removed, but the femoral canal was not plugged. The rate of revision and risk factors were evaluated by survivorship analysis. The overall cumulative revision rates were 11%, 14% and 19% after 8, 10, and 14 years respectively. When prophylactic antibiotics were not used, there was a significant risk of revision for infection. The risk of revision for mechanical loosening of both the acetabular and femoral components was significantly increased for men, younger patients, and if a snap-fit prosthesis with a 35 mm head and short wedged stem was used. The survivorship findings were validated by multivariate statistical analysis.