As knee implants become more common, it is important to study their potential health risks. We investigated cancer occurrence in a nationwide population-based cohort of 30,011 patients who underwent knee replacement surgery in Sweden from 1980 to 1994. Patients were followed from 1 year after the date of their surgery through December 31, 1995, accruing 122,616 person-years of observation. The average follow-up time was 4.3 years, with 2365 patients followed for 10 years or more. Overall cancer incidence was not elevated compared with the general population of Sweden (standardized incidence ratio [SIR] = 1.03; 95% confidence interval [CI] = 0.98-1.08). A reduced rate for all respiratory cancers (SIR = 0.73; 95% CI = 0.59-0.91) and for lung cancer (SIR = 0.73; 95% CI = 0.58-0.91) was found among both men and women. Elevated rates were found for prostate (SIR = 1.20; 95% CI = 1.06-1.34) and bone cancer (SIR = 6.00; 95% CI = 1.24-17.52) in men. The bone cancer excess was based on three observed cases, two of which occurred at a site unrelated to the implant and the site of the third tumor is unknown. Rates of connective tissue cancer and leukemia-lymphoma were not elevated significantly among knee implant recipients. Long-term follow-up (>or= 10 years) did not show a significant excess risk for all cancer (SIR = 1.10; 95% CI = 0.86-1.38) or for any site-specific cancer, including bone cancer, lymphoma, or leukemia. Subgroup analyses for patients with rheumatoid arthritis produced results similar to the overall results. This epidemiologic study of cancer risk among patients with knee implants is the largest to date. It provides evidence that the incidence of cancer among patients with knee implants is similar to that of the general population. Continued follow-up of this cohort is warranted to evaluate further potential long-term effects of these implants.