The aim of this study was to investigate the ability of 111In-labelled human polyclonal immunoglobulin G (111In-IgG) to localize bone and joint infections compared with 99Tcm-HMAPO-labelled leukocytes (99Tcm-WBC). Thirty-four patients routinely referred for investigation of bone and joint infections were studied. In all patients, a bone scan using 99Tcm-MDP was initially obtained. Subsequently, 99Tcm-WBC and 111In-IgG were simultaneously injected and images obtained at 30 min, 4 h and 24 h post-injection. Diagnostic accuracy was established by bacteriology of specimens obtained by needle aspiration and/or surgery, other imaging methods and clinical follow-up. The images were read by three experienced observers blinded to any other information; the clinical suspicion of infection and the diagnosis were established when two observers agreed. Infection was confirmed in 11 patients. The 99Tcm-WBC scans gave 8 true-positive, 5 false-positive, 18 true-negative and 3 false-negative results. With 111In-IgG, the figures were 7, 6, 17 and 4, respectively. The sensitivity, specificity and accuracy were 72.7%, 78.2% and 76.4% respectively for the labelled leukocytes and 63.6%, 73.9% and 70.6% respectively for 111In-IgG. There was greater agreement between the observers with 99Tcm-WBC than 111In-IgG. In this study, 111In-IgG was less sensitive and less specific than 99Tcm-WBC scintigraphy for the diagnosis of chronic infections, but these differences were not significant. Both tracers appear to be useful in the diagnosis of bone and joint infections. However, our results were less reliable for the diagnosis of an infected prosthesis.