Thirty-seven patients who were clinically suspected of having vertebral osteomyelitis were prospectively evaluated with magnetic resonance (MR), radiography, and radionuclide studies. These findings were correlated with the final clinical, microbiologic, or histologic diagnoses. Based on the results of these latter studies, 23 patients were believed to have osteomyelitis. MR examinations consisted of at least a sagittal image (TE = 30 msec, TR = 0.5 sec) and an image obtained at TE = 120 msec, TR = 2-3 sec. All patients underwent radiographic and MR examinations, 36 underwent technetium 99m-HDP bone scanning, and 20 patients underwent gallium 67 scanning. Nineteen patients underwent both bone and gallium scanning. The imaging studies were reviewed independently by investigators blinded to the final diagnoses. MR had a sensitivity of 96%, specificity of 92%, and accuracy of 94%. Combined gallium and bone scan studies (19 cases) had a sensitivity of 90%, specificity of 100%, and accuracy of 94%. Bone scans alone had a sensitivity of 90%, specificity of 78%, and accuracy of 86%. Plain radiographs had a sensitivity of 82%, specificity of 57%, and accuracy of 73%. The MR appearance of vertebral osteomyelitis in this study was characteristic, and MR was as accurate and sensitive as radionuclide scanning in the detection of osteomyelitis.