The records of 162 patients who underwent technetium-99m-tagged red blood cell scans for the localization of gastrointestinal hemorrhage were studied. Ninety-eight scans were read as positive, with bleeding sites determined by a radiologist. Forty-six patients had a definitive diagnosis made by other means. Tagged scans accurately localized the site of bleeding in only 52% of cases. Analysis of the value of these tagged scans in predicting a subsequent positive angiogram indicated that there were 14 positive and 12 negative angiograms of 26 positive scans. Of the nine patients with negative scans who underwent arteriography, the arteriogram was positive in five. These data suggest that tagged red blood cell scanning is a poor diagnostic technique for the localization of gastrointestinal bleeding, and its use as a screening tool before angiography is questionable.