Diagnosis of osteomyelitis of the foot in diabetic patients may be difficult because of the coexistence of chronic cellulitis, vascular insufficiency, and peripheral neuropathy. This study compared the diagnostic accuracies of plain films, bone scans, and MR imaging studies in diabetic patients with suspicion of osteomyelitis of the foot. Twenty-nine plain radiographs, 20 bone scans, and 30 MR studies were obtained in 24 patients. Twenty-nine bones from 14 patients were pathologically proved either positive (25 bones) or negative (four bones) for osteomyelitis. Another 15 bones (10 patients) studied with MR had no pathologic proof, but the bones healed with only local wound care and/or a short course of oral antibiotics. These patients had trauma, cellulitis, or unhealed ulcers. The sensitivity and specificity of plain films were both 75%. Bone scans had a very low specificity (100% false-positive rate). A negative bone scan should strongly exclude the probability of osteomyelitis. Unlike the findings in previous reports, MR had much higher sensitivity and specificity than bone scans in detecting osteomyelitis in diabetic patients. When the 10 patients without pathologic proof (those who presumably had neuroarthropathy, vascular insufficiency, and/or cellulitis) were included, the sensitivity and specificity of all three techniques decreased. Our experience with this small group of patients suggests that MR is a useful imaging technique for diagnosing osteomyelitis of the foot in diabetic patients.