Embolization represents a standard of care for gastrointestinal (GI) bleeding. However, there is limited data on its efficacy in the setting of positive radiologic or endoscopic imaging and negative angiography. Here, we assess outcomes, complications, and perioperative variables influencing efficacy of empiric embolization for GI bleeding. 55 patients underwent embolization for upper (UGIB) or lower gastrointestinal bleeding (LGIB) with absent contrast extravasation on angiography between January 2012 and December 2015. Data on comorbidities, bleed character, procedure technique, and outcomes were collected. The primary endpoint was clinical success defined as hemostasis without rebleeding, complication, or bleeding-related mortality during the observation period. Secondary endpoints were technical success, 30-day all-cause mortality, and complications. Differences were assessed with Student’s T-test and Chi-square or Fisher’s exact test. Predictors of outcomes were modeled with a forward stepwise regression. Median follow-up was 817 (IQR: 105-1483) days. 47 (85%) embolizations were performed for UGIB and 8 (15%) for LGIB. Gastroduodenal (31%) and left gastric (22%) arteries were most commonly embolized, with coils alone (68%) being the most common method. Technical success was 98.2%. Clinical success was 62% and 38% for UGIB and LGIB, respectively. Median time to rebleed was 4 (IQR: 2-9) days. 30-day all-cause mortality was 29%. No UGIBs developed complications whereas 2 LGIBs were complicated by ischemic colitis and perforated diverticulum. Pressor support (P = 0.02) and >5 units PRBCs transfused (P = 0.007) were associated with short-term (7 day) clinical failure, with pressor support independently predictive on multivariate analysis (OR: 0.20, 95% CI: 0.04-0.89; P = 0.04). Liver disease (P = 0.04) and coagulopathy (P = 0.04) were associated with overall clinical failure. Empiric embolization for UGIB with negative angiography is effective at decreasing rebleeding; though, it may be less successful in patients with baseline coagulopathy or liver disease. The low efficacy and potential complications associated with LGIB warrant careful consideration in this population.