Melanoma is characterized by a tendency to metastasize widely throughout the body and its relative affinity for gallium-67. Because of the ability of this nuclide to image tumor sites in numerous organ systems, it has been used to detect metastases in patients with malignant melanoma. The effectiveness of this technique, however, has been controversial. This article documents the retrospective analysis of results from 296 gallium-67 scintiscans from 222 patients with melanoma. Patients were placed in two groups. The low suspicion group (148 patients undergoing 191 scans) consisted of patients with no evidence of disease; the gallium scans were performed solely for screening purposes. There were 175 true-negative scans, nine true-positive scans (eight of the nine were positive only at the untreated primary site), three false-negative scans, and four false-positive scans. Therefore, in only one patient (0.5%) did a "screening" gallium scan reveal disease that was not expected. The high suspicion group (85 patients undergoing 105 scans) consisted of patients with established evidence of metastatic disease; the gallium scan was performed to confirm those findings and to search for involvement of other organ systems. Of these scans, ten were true-negative, 73 true-positive, 21 false-negative, and one false-positive. In this group the 20% false-negative results indicate that gallium scanning is considerably less sensitive than the combination of clinical and standard radiographic assessment. It was concluded that gallium-67 scintiscanning of patients with melanoma, whether for screening or evaluation of patients with known metastases, provides little information that affects clinical staging or therapeutic design. Therefore, the technique is of limited value for routinely investigating the extent of disease.