Objective This study was designed to evaluate the role of a negative computed tomography angiogram (CTA) in patients who present with gastrointestinal (GI) hemorrhage. Methods A review of all patients who had CTAs for GI hemorrhage over an 8-year period from January 2005 to December 2012 was performed. Data for patient demographics, location of hemorrhage, hemodynamic stability, and details of angiograms and/or the embolization procedure were obtained from the CRIS/PACS database, interventional radiology database, secure electronic medical records, and patient's clinical notes. Results A total of 180 patients had 202 CTAs during the 8-year period: 87 CTAs were performed for upper GI hemorrhage (18 positive for active bleeding, 69 negative) and 115 for lower GI hemorrhage (37 positive for active bleeding, 78 negative); 58.7 % (37/63) of patients with upper GI bleed and 77.4 % (48/62) of patients with lower GI bleed who had an initial negative CTA did not rebleed without the need for radiological or surgical intervention. This difference was statistically significant (p = 0.04). The relative risk of rebleeding, following a negative CTA, in lower GI bleeding versus upper GI bleeding patients is 0.55 (95 % confidence interval 0.32-0.95). Conclusions Patients with upper GI bleed who had negative CTAs usually require further intervention to stop the bleeding. In contrast, most patients presenting with lower GI hemorrhage who had a negative first CTA were less likely to rebleed.