Background: The blood urea nitrogen to creatinine ratio (BUN/CREAT) is believed to reliably discriminate upper gastrointestinal bleeding (UGIB) from lower gastrointestinal bleeding (LGIB). However, studies evaluating subsets of bleeders in whom this ratio may have real diagnostic utility are lacking. Methods: Over a 50-month period, all patients evaluated for UGIB and LGIB by our gastroenterology consultative service had demographic, clinical, and laboratory findings recorded on admission. Endoscopic evaluation was performed in most patients for diagnosis. Results: A total of 790 patients with UGIB and 162 with LGIB were studied. Peptic ulcer disease (57 %) and esophageal varices (10%) were the most common causes of UGIB, whereas diverticulosis was etiologic in 54% of LGIB episodes. The mean (± SD) BUN/CREAT ratio was significantly higher in UGIB than LGIB (22.5 ± 11.5 vs 15.9 ± 8.2; p = 0.0001). When comparing UG1B patients without hematemesis or diagnostic nasogastric aspirate to patients with LGIB, significant differences were found for patients with melena (20.1 ± 8.4 vs 15.9; p = 0.001) but not hematochezia (18.6 ± 9.1 vs 15.9; p = 0.12), and overlap was great. Using a ratio of ≤ 33, the sensitivity and specificity for LGIB was 96 and 17%, respectively. There was a significant correlation of transfusion requirements and admission hematocrit to this ratio, whereas admission vital signs were not found to correlate significantly. Conclusions: Although the BUN/CREAT ratio is higher in UGIB compared with LGIB, the degree of overlap, especially in patients without hematemesis, suggests this value to have poor discriminatory ability. The degree of elevation correlates best with transfusion requirements rather than admission vital signs.