Between January and July 1998, we conducted a prospective study to compare Tc-99m-labelled antigranulocyte monoclonal antibody fragment Fab' (LEUKOSCAN) scintigraphy versus Tc-99m-hexamethylpropyleneamine oxime (Tc-99m-HMPAO)-labelled leukocyte scintigraphy (HMPAO-LS) for the diagnosis of unselected patients with bone and joint infection. Twenty-three patients (16 men and 7 women; mean age, 67 years) with suspected bone infection were explored successively with bone scintigraphy, HMPAO-LS and LEUKOSCAN scintigraphy. Thirty-two foci were studied (diabetic foot = 11, prosthetic material = 8, joint disease = 4, others = diagnosed in 18 cases, eight on the basis of bacteriological and histological examination of surgical or puncture specimens, with or without radiographic signs, and 10 on the basis of clinical course and radiographic findings. Overall sensitivity, specificity and accuracy were 86%, 72% and 78%, respectively, for LEUKOSCAN scintigraphy (12 true positives (TP), 13 true negatives (TN), 5 false positives (FP), 2 false negatives (FN)), 93%, 100% and 96%, respectively, for HMPAO-LS (13TP, 18TN, 0FP, 1FN), and 100%, 17% and 53.3%, respectively, for bone scintigraphy. In this small series, LEUKOSCAN scintigraphy was found to be less specific for the diagnosis of osteomyelitis than HMPAO-LS. In addition, the interpretation of LEUKOSCAN scintigraphy is more difficult than HMPAO-LS for the diagnosis of bone infection in the diabetic foot, and would appear to be less discriminating for differentiating soft tissue infection from osteitis in the case of plantar perforating ulcers.