Objectives: Upper gastrointestinal bleeding (UGIB) is a critical medical emergency with a potentially fatal outcome. Early risk stratification is essential for determining the need for urgent interventions. Current guidelines recommend the use of risk-stratification models, among which the ABC score was recently developed. We aimed to validate the performance of the ABC score in comparison with the well-established Glasgow–Blatchford score (GBS), AIMS65, and pre-endoscopic Rockall score (PreRS). Methods: This retrospective single-center study included adult patients (≥18 years) who presented to the emergency department of Incheon St. Mary’s Hospital with non-variceal UGIB between March 2019 and June 2022. The primary outcome was 30-day all-cause mortality. Secondary outcomes included a composite endpoint of hemostatic intervention (endoscopic, interventional radiologic, or surgical), hypotension (systolic blood pressure <90 mm Hg after 2 h), vasopressor use after 2 h, and rebleeding within 7 days. The predictive performance of the GBS, AIMS65, PreRS, and ABC scores was assessed using the area under the receiver operating characteristic curve (AUROC). Results: Of the 1597 enrolled patients, 116 (7.3%) died within 30 days. The ABC score demonstrated the highest performance (AUROC: 0.806; 95% confidence interval: 0.766–0.845) at predicting 30-day mortality, followed by the PreRS (0.734), GBS (0.679), and AIMS65 (0.558). The GBS had the highest AUROC (0.708), followed by the ABC (0.651), PreRS (0.626), and AIMS65 (0.529). Conclusions The ABC score outperformed conventional risk models at predicting the 30-day mortality among patients with non-variceal UGIB. However, its predictive power for the need for intervention was inferior to that of the GBS.