Upper gastrointestinal bleeding (UGIB) presents a high incidence in an emergency department (ED) and requires careful evaluation of the patient’s risk level to ensure optimal management. The primary aim of this study was to externally validate and compare the performance of the Rockall score, Glasgow-Blatchford score (GBS), modified GBS and AIMS65 score to predict death and the need for an intervention among patients with UGIB. This was a cross-sectional observational study of patients consulting the ED of a Swiss tertiary care hospital with UGIB. Primary outcomes were the inhospital need for an intervention, including transfusion, or an endoscopic procedure or surgery or inhospital death. The secondary outcome was inhospital death. We included 1521 patients with UGIB, median age, 68 (52–81) years; 940 (62%) were men. Melena or hematemesis were the most common complaints in 1020 (73%) patients. Among 422 (28%) patients who needed an intervention or died, 76 (5%) died in the hospital. Accuracy of the scoring systems assessed by receiver operating characteristic curves showed that the Glasgow-Blatchford bleeding and modified GBSs had the highest discriminatory capacity to determine inhospital death or the need of an intervention [AUC, 0.77 (95% CI, 0.75–0.80) and 0.78 (95% CI, 0.76–0.81), respectively]. AIMS65 and the pre-endoscopic Rockall score showed a lower discrimination [AUC, 0.68 (95% CI, 0.66–0.71) and 0.65 (95% CI, 0.62–0.68), respectively]. For a GBS of 0, only one patient (0.8%) needed an endoscopic intervention. The modified Glasgow-Blatchford and Glasgow-Blatchford bleeding scores appear to be the most accurate scores to predict the need for intervention or inhospital death.