OBJECTIVE Technetium-99m-labeled red blood cell scans (Tc99m RBC scan) are recommended to confirm gastrointestinal (GI) bleeding. It is controversial whether these scans are sufficient to localize the site of bleeding. This study evaluated the efficacy of RBC scans in confirming and localizing GI bleeding. Our hypothesis was that these scans were effective in localizing GI bleeding if positive within the continuous phase of imaging. METHODS Tc99m RBC scans were performed on a total of 80 patients over a four-year period to localize GI bleeding (59 male, 21 female; age range 6-88 (mean, 48) years). Films of 75 of the 80 patients were reread by a nuclear medicine physician who was blinded to the original reading and identity and history of the patient. Results of scans were compared with confirmatory studies. RESULTS A total of 21 patients had positive scans (28 percent). Of these, the site of bleeding in 16 of 21 patients (76 percent) was confirmed by angiography (4/16), endoscopy (10/16), surgery (10/16), or a combination of these. In 14 of the 16 confirmed studies (88 percent), RBC scan correctly localized site of bleeding by our rigid definition. In six patients (4 not confirmed, 2 erroneously localized), scans were positive only at greater than 15 hours. Ten of the 14 correctly localized studies and none of the incorrectly localized studies were positive in the continuous phase of imaging. CONCLUSIONS Tc99m RBC scan is effective in localizing GI bleeding when positive within the continuous phase of imaging. In this population supplemental angiography or endoscopy for the purpose of localization would seem unnecessary.