Numerous clinical studies have questioned the ability of radionuclide scans to differentiate septic from aseptic joint inflammation. A clinical study may not be able to document an underlying disease process or duration of infection and, thus, may make conclusions about the accuracy of scan interpretations open to debate. In this study, the Dumonde- Glynn model of antigen-induced arthritis in rabbits was used as the experimental model to study technetium and gallium scans in Staphylococcus aureus infection of arthritic and normal joints. Galium scans were negative in normal rabbits, usually negative in antigen-induced arthritis, but positive in septic arthritis. The bone scan was usually negative in early infection but positive in late septic arthritis, a finding reflecting greater penetration of bacteria into subchondral bone because of the underlying inflammatory process.