Magnetic resonance (MR) images and computed tomographic (CT) scans of histologically characterized soft-tissue masses of the locomotor system in 35 patients were compared for image contrast, demonstration of bone destruction, and display of extent and anatomic relationships of the masses. Subjective criteria for predicting malignancy were tested. T1 measurements were made in a few cases. Intensities of masses relative to those of fat and muscle in spin-echo T1-weighted and highly T2-weighted images were evaluated for correlation with tissue type. Subjective value of using coronal and sagittal images was assessed. Because of its superior inherent image contrast and its ability to provide direct sagittal and coronal images, MR was better than CT in demonstrating size and extent of most tumors and their relationships to vascular and nonvascular structures. However, bone destruction was more difficult to see with MR. Except for fatty tumors, MR was not helpful in identifying tissue type. Subjective criteria were of limited value in distinguishing benign from malignant lesions. Moreover, there is currently no credible evidence that T1 or T2 measurements are helpful in this regard. Study results suggest that MR is superior to CT in evaluating soft-tissue masses of the locomotor system. If an MR examination is performed, CT may not be necessary in certain cases unless bone involvement is suspected.