A significant proportion of patients with acute, non-variceal UGIH are at low-risk for adverse outcomes (rebleeding or death). The Rockall score is a valid predictive index for stratifying adverse outcome risk based on clinical (age, co-morbidity, vital signs) and endoscopic (source of bleed, stigmata of recent hemorrhage) criteria. Prolonged hospitalization of patients with low-risk Rockall scores may not be justified. Purposes were to: 1) identify the incidence of low-risk, non-variceal UGIH admissions to a tertiary care referral center during 1997 and 1998, 2) determine mean hospital length of stay (HLOS) for this cohort, and 3) determine the incidence of adverse outcomes. Methods: Cases were identified using ICD-9 codes for discharge diagnosis. Medical records were abstracted using a standardized form. SAS was used for data management/analysis. Results: see TABLE. Of 294 cases identified, 118 were excluded (lower GIH, no endoscopy performed, in-hospital bleed, transfer), leaving 175 cases of non-variceal UGIH for analysis. 95/175 (54%) were men. Mean age = 62 yrs. 22% used NSAIDS, and 30% ASA. 117 (67% of all, and 47% of low-risk) were admitted to an ICU or monitored bed. Most common diagnoses were GU in 40 (23%) and DU in 23 (13%). Among Rockall low-risk patients, adverse outcomes were as follows: 8/70 (11%) received transfusions after endoscopy, yet only 1/70 (1.4%) required repeat endoscopy and surgery, 3/70 (4.2%) required hospital re-admission for re-bleeding within 30 days of discharge, and 0/70 (0%) died. Conclusions: Incidence of Rockall low-risk, non-variceal UGIH admissions is high, while adverse outcomes are rare. Utilization of healthcare resources, including level of care and HLOS, appears excessive and unwarranted for this cohort.