BACKGROUND/AIMS: The Glasgow Blatchford Bleeding Score (GBS) has been developed to assess the need for treatment in patients with acute upper gastrointestinal hemorrhage (UGIH) presenting at emergency departments (EDs). We aimed (a) to determine the validity of the GBS and Rockall scoring systems for prediction of need for treatment and (b) to identify the optimal cut-off value of the GBS. METHODS: We carried out a population-based, prospective multicenter study of 520 consecutive patients presenting with acute UGIH at EDs of three hospitals. The accuracy of GBS and Rockall scores in predicting the need for treatment (i.e. endoscopic, surgical, or radiological intervention and blood transfusion) was analyzed using receiver operating characteristic curves. RESULTS: Receiver operating characteristic curve analysis showed that the GBS had a good discriminative ability to determine the need for treatment in patients with acute UGIH (area under the curve: 0.88; 95% confidence interval: 0.85-0.91). The GBS was superior to both the clinical Rockall and the full Rockall score in predicting the need for treatment (area under the curve: 0.86 vs. 0.70 vs. 0.77). At a cut-off value of up to 2, the GBS had the optimal combination of sensitivity (99.4%) and specificity (42.4%). CONCLUSION: The GBS is superior compared with both Rockall scores in predicting the need for treatment in patients with suspected acute UGIH presenting at EDs in the Netherlands. Patients with a GBS of 2 or less form a subgroup of low-risk patients. These low-risk patients are eligible for outpatient management, which might reduce hospital admissions and healthcare costs.