The aim of this study was to assess the accuracy of technetium 99m-labeled red cell scintigraphy in localizing the site of lower gastrointestinal bleeding. The outcome of 203 patients undergoing technetium 99m-labeled red cell scintigraphy was reviewed, and the scan result was compared with the true site of bleeding. The true site of bleeding was determined by other methods including angiography and surgical pathology. Fifty-two scans (26%) were positive and indicated a specific site of bleeding. A definitive bleeding site was identified in 22 patients by other means and correlated with the technetium scan in only 9 cases. The nuclear scan was incorrect in the remaining 13 cases, implying a localization error of 25% (13 of 52). A subgroup of 19 patients with a positive scan underwent a surgical procedure directed by the nuclear scan. Eight of these 12 patients had incorrect surgical procedures based upon findings of more definitive tests, indicating a surgical error of 42% (8 of 19). We conclude that the technetium 99m-labeled red cell scan's ability to accurately localize the site of lower gastrointestinal bleeding is limited. Furthermore, performing a surgical procedure that relies exclusively on localization by red cell scintigraphy will produce an undesirable result in at least 42% of patients.