A cross-sectional, community-based survey of a random sample of 1750 of 242,311 Medicare recipients was performed. The patients were at least sixty-five years old and had had a primary or revision knee replacement (either unilaterally or bilaterally) between 1985 and 1989. Three samples were surveyed separately: a national sample (to reflect the United States as a whole) and samples from Indiana and the western part of Pennsylvania (sites chosen for convenience to assess the validity of the findings for the national sample on a regional level). Each sample was stratified by race, age, residence (urban or rural), and the year of the procedure. Valid and reliable questionnaires were used to elicit the participants' assessments of pain, physical function, and satisfaction two to seven years after the knee replacement. Of the 1486 patients who were eligible for inclusion in the survey, 1193 (80.3 per cent) responded. The mean age of the respondents was 72.6 years. Eight hundred and forty-nine respondents (71.2 per cent) were white, and 849 (71.2 per cent) were women. The participants reported that they had little or no pain in the knee at the time of the survey, regardless of the age at the time of the knee replacement, the body-mass index, or the length of time since the knee replacement. After adjustment for potential confounding variables, predictors of better physical function after the replacement were an absence of problems with the contralateral knee, primary knee replacement (rather than revision) (Indiana sample only), and a lower body-mass index (Indiana and western Pennsylvania samples). Four hundred and fifteen (85.2 per cent) of the 487 patients in the national sample were satisfied with the result of the knee replacement. In what we believe to be the first community-based study of the outcome of knee replacement, patients reported having significant (p = 0.0001) and persistent relief of pain, improved physical function, and satisfaction with the result two to seven years postoperatively. The findings of the present study suggest that age and obesity do not have a negative impact on patient-relevant outcomes (pain and physical function). Dissemination of these findings has the potential to increase appropriate referrals for knee replacement and thereby reduce the pain and functional disability due to osteoarthrosis of the knee.