Pelvic pain in cancer patients can result from several causes. The most appropriate choice of imaging techniques for evaluating such patients has not been established. We evaluated 27 cancer patients with pelvic pain by using radionuclide bone scintigraphy (24 patients), abdominal CT (27 patients), and pelvic MR imaging (27 patients) and used the correlation between symptoms and imaging findings to compare these imaging methods. The study population included 11 patients with Ewing sarcoma, six with other sarcomas, five with colorectal cancers, and five with other tumors. All patients had pelvic pain, and eight had pain radiating to a leg. Twenty-three patients had soft-tissue masses, and 19 had bone metastases; 16 had both. Findings on bone scans explained the symptoms in 17 (71%) of 24 patients, findings on CT in 23 (85%) of 27 patients, and findings on MR imaging in 25 (93%) of 27 patients. The difference between bone scanning and CT or MR was statistically significant (p less than .05); however, the difference between CT and MR imaging was not significant (p greater than .05). MR imaging detected 41 (98%) of 42 relevant lesions, whereas CT detected 31 (74%) of 42, and bone scanning 17 (44%) of 39. We conclude that MR is superior to either bone scanning or CT in the initial evaluation of pelvic pain in cancer patients. Such information can be important in directing the treatment of these patients.