OBJECTIVE Acute non-variceal gastrointestinal bleeding (NVGIB) is one of the most common medical emergencies, leading to significant morbidity and mortality without proper management. This study was to analyze the causes of NVGIB and to evaluate the safety, efficacy, and feasibility of transcatheter arterial embolization (TAE) for the treatment of NVGIB. METHODS From November 2012 to October 2018, 158 patients with NVGIB underwent digital subtraction angiography, and TAE was performed for confirmed gastrointestinal bleeding. Patient characteristics, cause of bleeding, angiographic findings, technical and clinical success rates, complication rates, and outcomes were retrospectively analyzed. RESULTS Bleeding was confirmed in 71.5% (113/158) of performed angiographies, and 68 patients had visible contrast extravasation on angiography, with the other 45 patients having indirect signs of bleeding. Among the 113 patients with confirmed gastrointestinal bleeding, TAE was technically successful in 111 patients (98.2%). The mean procedure time required for TAE was 116 ± 44 min (ranging from 50 to 225 min). The primary total clinical success rate of TAE was 84.7% (94/111). The primary clinical success rates of TAE for vascular abnormality, neoplastic disease, and iatrogenic condition were 84.5% (49/58), 84.1% (37/44), and 88.9% (8/9), respectively. Intestinal necrosis and perforation were found in two patients after TAE. CONCLUSIONS The causes of NVGIB are complex and the onset, location, risk, and clinical presentations are variable. NVGIB can be generally divided into three types: vascular abnormality, neoplastic disease, and iatrogenic condition. TAE is a safe, effective, and fast procedure in the management of gastrointestinal bleeding.