The purpose of this study was to determine differences in the use of total knee arthroplasty for osteoarthritis among black and white older Americans and to determine whether clinical and economic factors contribute to these differences. Data on black and white Americans aged 65 years and older were gleaned from national Medicare hospitalization records for 1980 through 1988 and the Medicare enrollment file for 1987, the First National Health and Nutrition Examination Survey of 1971 through 1975, the National Hospital Discharge Survey of 1979 through 1988, the 1980 through 1988 US census counts and estimates, and 1980 census economic data. Information gathered includes prevalence of osteoarthritis of the knee among blacks and whites; rate of total knee arthroplasty use in the United States by race, sex, and age of subjects at the time of operations; use of competing procedures; and economic factors. Results showed that blacks were less often treated with total knee arthroplasty than were whites (white-to-black rate ratios = 3.0 to 5.1 for men and 1.5 to 2.0 for women) but had nonsignificantly higher rates of clinical osteoarthritis of the knee (white-to-black rate ratios = 0.39 for men and 0.78 for women). This racial difference in total knee arthroplasty rates was consistent across income levels and was unexplained by black patients' having operations at earlier ages or using competing procedures. Furthermore, the discrepancy occurred even among Medicaid-eligible Medicare recipients, who had no direct economic disincentive for surgery. These findings suggest that even though elderly blacks have higher rates of knee osteoarthritis, they do not receive total knee arthroplasty as often as do elderly whites. This discrepancy does not appear to have an economic explanation. Even if a comprehensive care system were in place to remove economic barriers, inequality in the use of this procedure would likely persist. Future studies should assess the nonclinical and noneconomic causes of these differences.