Published by the BMJ Publishing Group Limited. For permission to use (where not already granted under a licence) please go to http://group.bmj.com/group/rights-licensing/permissions. OBJECTIVE: To examine whether racial disparities in usage and outcomes of total knee and total hip arthroplasty (TKA and THA) have declined over time. METHODS: We used data from the US Medicare Program (MedPAR data) for years 1991-2008 to identify four separate cohorts of patients (primary TKA, revision TKA, primary THA, revision THA). For each cohort, we calculated standardised arthroplasty usage rates for Caucasian and African-American Medicare beneficiaries for each calendar year, and examined changes in disparities over time. We examined unadjusted and adjusted outcomes (30-day readmission rate, discharge disposition etc.) for Caucasians and African-Americans, and whether disparities decreased over time. RESULTS: In 1991, the use of primary TKA was 36% lower for African-Americans compared with Caucasians (20.6 per 10,000 for African-Americans; 32.1 per 10,000 for Caucasians; p < 0.0001); in 2008, usage of primary TKA was 40% lower for African-Americans (41.5 per 10,000 for African-Americans; 68.8 per 10,000 for Caucasians; p < 0.0001) with similar findings for the other cohorts. Black-White disparities in 30-day hospital readmission increased significantly from 1991-2008 among three patient cohorts. For example in 1991 30-day readmission rates for African-Americans receiving primary TKA were 6% higher than for Caucasians; by 2008 readmission rates for African-Americans were 24% higher (p < 0.05 for change in disparity). Similarly, black-white disparities in the proportion of patients discharged to home after surgery increased across the study period for all cohorts (p < 0.05). CONCLUSIONS: In an 18-year analysis of US Medicare data, we found little evidence of declines in racial disparities for joint arthroplasty usage or outcomes.