BACKGROUND: Because local delivery of drugs induces high concentrations, it could be helpful to apply these delivery systems to the treatment of septic arthritis by antibiotics. Thus, a gentamicin-loaded polymer was tested in a rabbit model of Staphylococcus aureus septic arthritis. METHODS: Thirty New Zealand rabbits were split into five groups: A: infection only; B: infection and systemic gentamicin treatment; C: infection and unloaded polymer and systemic gentamicin treatment; D: infection and gentamicin-loaded polymer only; and E: no infection and unloaded polymer. After inducing nonlethal septic arthritis in the knee joint by injecting 10 colony-forming units (CFUs) of a strain of methicillin-sensitive S aureus in groups A, B, C, and D, rabbits were housed for 15 days, and then the joint capsules were removed and the remaining bacteria were counted. Bacterial load was expressed in CFUs per gram of synovial tissue. In group E, capsules were removed, and a pathologic examination was done. RESULTS: At day 15, the bacterial load was 6 × 10, 2 × 10, 1.8 × 10, and 7 × 10 CFU/g of tissue for groups A, B, C, and D, respectively. Compared with the mean of groups A, B, and C, the bacterial load of group D was 4.94 units of log10 CFU/g lower than that of these groups. The bacterial load of group D was statistically significantly lower than that of the other three groups. Noticeably, two animals of group D had a nil bacterial count. In group E animals, a minimal foreign body reaction was observed around the polymer. DISCUSSION: Gentamicin-containing microparticles were more efficient in reducing bacterial load than systemic injections of gentamicin and thus have an interesting role to play in the treatment of human arthritis. However, inserting microparticles in joints is not easy, and hydrogels might be a good alternative approach.