The authors analyzed the 5-year and 9-year survival in 134 of 165 patients who underwent total knee arthroplasties from 1989 to 1996 in our department. Patients were followed until December 31, 1998, or until the time of death. Diagnoses were rheumatoid arthritis in 81 patients (132 knees) and osteoarthritis in 53 patients (79 knees). The survival of the patients was compared to that of the age- and sex-adjusted general population. Kaplan-Meier survival curves were constructed. Twenty-two patients in the study died before the end of the follow-up. The cumulative 5-year patient survival was 88.7%, and 9-year patient survival was 64.4% for total knee arthroplasty patients. The standardized mortality ratio was 0.11 (95% confidence interval: 0.02-0.40) for the patients with osteoarthritis, and 0.81 (95% confidence interval: 0.52-1.25) for the patients with rheumatoid arthritis. The Cox proportional hazards model showed that the factors of male sex and rheumatoid arthritis were related to a higher mortality rate in the total knee arthroplasty group.