Background <p>Gastrointestinal bleeding is a major cause of morbidity and mortality, this study aimed to validate the performance of the new ABC score in predicting 30-day mortality in upper gastrointestinal bleeding (UGIB) and lower gastrointestinal bleeding (LGIB) in a cohort of Tanzanian patients and compare it to existing pre-endoscopy risk scores.</p> Methods <p>This retrospective single-center study at a tertiary referral hospital in Tanzania included adult patients presenting with gastrointestinal bleeding between June 2022 and December 2023. The area under receiver operating characteristic curve (AUROC) was calculated for the ABC score 30-day mortality prediction using SPSS software. In addition, the top risk prediction existing scores (the AIMS65, pre-endoscopic Rockall score, Glasgow-Blatchford score (GBS) and, the Oakland score) were evaluated and compared to the ABC score using DeLong tests.</p> Results <p>Among the 98 included patients the overall 30-day mortality was 31.6%. The ABC score had a good performance in predicting 30-day mortality (AUROC: 0.818, 95% CI: 0.732–0.903). The ABC score’s performance was superior to GBS (<i>p</i> = 0.015) and Oakland scores (<i>p</i> = 0.008), while comparable to AIMS65 (<i>p</i> = 0.083) and Rockall scores (<i>p</i> = 0.032). Notably, the optimal ABC cutoff in our population was ≥ 4.5 points, differing from the international standard of ≥ 8 points.</p> Conclusions <p>Our study provides the first African validation of the ABC score, confirming a good predictive performance in Tanzanian patients while identifying the need for population-specific cutoff calibration. The ABC score represents a valuable risk stratification tool for both upper and lower gastrointestinal bleeding in resource-limited settings.</p>