Experimental bacterial infection following implant arthroplasties was investigated in rabbits. Bone cement and a stainless steel head and stem prosthesis were inserted after reaming of the femoral neck and shaft. Measured doses of Staphylococcus aureus were injected either into the femoral medullary cavity or intravenously. Intravenous challenge required high inocula to establish arthroplasty infection, whereas infection around the prosthesis was consistently established after inoculation of 10(3) bacteria into the femoral medulla. The erythrocyte sedimentation rate (ESR) proved the most reliable clinical test of infection. Hematogenous infection was difficult to reproduce. Three weeks after operation, the arthroplasty was as resistant as the normal hip. Antibiotics were administered in doses equivalent to doses used in the treatment of infections in humans. When gentamicin-impregnated cement was used, 60 times the number of organisms were required to establish infection. Flucloxacillin and Imipenem gave similar protection. Rifampicin gave protection to a level exceeding the lethal dose of organisms for the model.