Introduction Upper gastrointestinal bleeding (UGIB) is caused by variceal lesions related to portal hypertension (PHT), or non-variceal lesions (NVL). Some patients may present with both PHT and NVL. We aimed to study the outcomes and possible interaction between the two conditions with adjustment for age, sex, smoking and alcohol. Methods Patients presenting with UGIB were classified according to the presence or absence of PHT or NVL, or both. PHT included varices in the oesophagus or stomach and gastropathy or duodenopathy. Other lesions were considered as non-variceal, such as erosive oesophagitis, peptic ulcers, erosive gastritis or duodenitis, etc. Logistic regression was used to assess PHT and NVL as predictive factors for UGIB outcomes, adjusting for demographic variables and testing for an interaction between PHT and NVL. Results Between 2008–2016, the following subgroups entered the analysis: No PHT/NVL (n=595, 56% males, median age 64 years); NVL only (n=1556, 63% males, age 67); PHT only (n=187, 63% males, age 56); and PHT+NVL (n=106, 65% males, age 58). The results of the logistic regression analysis are shown in table 1. Admission and 30-day mortality: Both PHT and NVL were independently predictive, with PHT having the stronger effect and with no significant interaction (table 1). For transfusion, PHT was strongly predictive but there was a highly significant interaction (P Conclusions There is a mismatch between the transfusion needs and clinical outcome in patients with both PHT and NVL, which may indicate that the bleeding is in some cases attributable to the NVL alone while the clinical outcome is related to the combined risk factors.