Experimental studies of osteomyelitis have been hampered by methodologic problems. For inducing experimental anaerobic osteomyelitis, the medullary cavity of the proximal tibial metaphysis of five New Zealand white rabbits was excavated and filled with a polyvinyl alcohol sponge. On the right side, 1 ml of a suspension containing Bacteroides fragilis (10(7) colony-forming units/ml) was injected, while on the left side, Ringer's solution was used as a control substance. Within five weeks, all animals developed clinical signs of osteomyelitis. The findings were verified by roentgenograms and bone scans. There was a significant rise in titers against the inoculated B. fragilis strain in every rabbit, and when killed at 18 weeks after the bacterial inoculation cultures containing the inoculated strain were obtained from all animals. No other bacterial strains were isolated. With one exception, cultures were positive in samples from both the inoculated and the control side, indicating hematogenous seeding. In one animal, the cultures from the inoculated side were negative, whereas on the control side, there was significant growth of the inoculated strain. Histologic examination of the infection sites showed low-grade chronic osteitis. In contrast to previous studies, which have indicated severe difficulty in obtaining a single-strain anaerobic osteomyelitis, the present method gives a high infection rate with reproducible immunologic, roentgenographic, and histologic reactions.