OBJECTIVES: Previous studies have found higher mortality rates among inpatients (IPs) compared with new admissions (outpatients, OPs) with acute upper gastrointestinal bleeding (AUGIB), but no studies have investigated the cause for this. The objective of this study was to determine whether the difference in outcomes between IPs and OPs with AUGIB can be explained by differences in baseline characteristics, bleeding severity, or processes of care. METHODS: Data were collected from 6,657 presentations with all-cause AUGIB from 212 UK hospitals as part of a nationwide audit. RESULTS: IPs were older (77 vs. 65 years, P<0.001), had greater comorbidity, and presented with more severe bleeding. There was no difference in median time to endoscopy (24 vs. 24 h, P=0.67) or receipt of endotherapy (19% vs. 17%, P=0.29). IPs had an odds of mortality 4.8 times that of OPs (26% vs. 7%; odds ratio (OR) 4.8, 95% confidence interval (CI) 3.9-5.8); after adjusting for baseline characteristics, this fell by 24% to 3.3 (95% CI 3.2-4.9) and after adjusting for bleeding severity alone to 4.0 (95% CI 3.2-4.9); adjusting for care processes had minimal effect. IPs had more than a twofold increased odds of rebleeding (20% vs. 12%; OR 2.1, 95% CI 1.7-2.5); adjusting for both baseline characteristics and severity of bleeding reduced this by 50% (OR 1.4, 95% CI 1.3-2.4), but process of care had no additional impact. CONCLUSIONS: IPs present with both higher baseline risks and more severe bleeding. These differences in baseline characteristics explain some but not all of the greater mortality of IPs with AUGIB.