The diagnostic accuracy for imaging infection with a technetium-99m-labeled antigranulocyte Fab' fragment (LeukoScan) was prospectively examined in a multicenter study. Scintigraphy was performed in 53 patients at 1 to 6 hours and at 24 hours after injection of the labeled antibody fragment. Thirty-nine sites of infection were detected and confirmed by histologic study, cytologic study, other imaging procedures, or by followup. Thirty-eight of the 53 patients also were studied with technetium-99m-Exametazim or indium-111-oxine labeled leukocytes within 1 week of the LeukoScan study. In 21 patients with 25 osteomyelitic lesions, LeukoScan recognized 13 of the lesions as being true positive ones, 10 as being true negative ones, and 2 as being false negative ones, whereas the leukocyte scan showed 9 true positive results, 5 true negative results, and 2 false negative ones. Sensitivity specificity, and diagnostic accuracy of LeukoScan were 90.0 %, 84.6 %, and 87.9 %; and with autologous leukocyte scintigraphy were 83.9%, 76.5%, and 81.3%, respectively. The sensitivity of LeukoScan was independent of the amount of the labeled antibody injected (0.1 - < 0.5 mg, 96.2%; 0.5 - < 0.9 mg, 80.0%; 0.9 - 1.0 mg, 77.8%). False positive lesions were detected in a periprosthetic calcification, a frontal hyperostosis, and 2 periprosthetic hips that had loosened. Human antimouse antibody could not be detected in any of the 13 patients tested 1 or 3 months after injection. LeukoScan is suitable for imaging infectious lesions and may have diagnostic advantages compared with autologous leukocyte scintigraphy.