BACKGROUND The implications and outcomes of patients with an uncleared fundal pool of blood found at emergent upper endoscopy are not well described. METHODS We reviewed the records of 484 consecutive patients who presented over a 12-month period to our medical center with acute upper gastrointestinal hemorrhage. All patients underwent upper endoscopy within 24 hours of their initial presentation. Patients with an uncleared fundal pool of blood at initial endoscopy were included in this study, and their findings and outcomes were compared with a randomly selected subgroup of these same patients who did not have residual gastric blood. RESULTS Sixty-one patients (13%) had uncleared fundal pools despite gastric lavage and patient positioning. Findings on initial endoscopy included esophageal varices in 29 (47%), gastric ulcer in 12 (20%), portal hypertensive gastropathy in 5 (8%), Mallory-Weiss tear in 5 (8%), duodenal ulcer in 5 (8%), gastric varices in 4 (7%), Dieulafoy's lesion in 2 (3%), and other in 7 (11%). Twelve of these 61 patients had multiple findings and 4 (7%) had no lesion identified. Thirty-two of the 61 patients (52%) had at least one follow-up endoscopy, with new fundal lesions identified in 13 (41%): portal hypertensive gastropathy in 8, gastric ulcer in 2, gastric varices in 2, and leiomyoma in 1. Of these 13 new findings, 5 (38%) were judged significant either by the presence of active bleeding or stigmata of recent hemorrhage. Of the 4 patients with no identifiable lesion on initial endoscopy, 3 had a follow-up endoscopy and 2 were found to have a significant new finding in the fundus. The control group had a statistically significant lower percentage of endoscopic findings related to portal hypertension. Recurrent bleeding during the index hospitalization occurred in 54% of the patients with uncleared fundal pools versus 11% of the control group (0 < 0.01). Length of stay, number of units of blood transfused, need for emergent surgery for bleeding, as well as overall and bleeding-related mortality were all significantly greater in the patients with the uncleared fundal pool than in the control patients. CONCLUSIONS The inability to clear a fundal pool of blood at emergent upper endoscopy is associated with significant morbidity and mortality. Further, new fundal lesions can be identified in 41% of patients on follow-up examination, with many being clinically significant. These data support the importance of clearing a fundal pool in patients undergoing endoscopy for upper gastrointestinal bleeding.