To analyze the possible differential features of upper gastrointestinal (GI) bleeding due to Dieulafoy's lesion (DL) compared with bleeding due to gastroduodenal ulcer.We evaluated variables associated with patient characteristics, hemorrhagic characteristics and clinical severity in 31 cases of bleeding secondary to DL and 31 cases of gastroduodenal ulcer bleeding. Univariate and multivariate statistical analysis were performed.The comorbidity rate was 80% in the DL group and 42% in the ulcer group (OR = 5.54; 95%CI, 1.5-20.7; p0.0004). Lesion location was gastric in 87% of DL versus 39% of ulcers (OR = 10.7; 95%CI, 2.6-47.6; p0.0003). More than one gastroscopy was required for diagnosis in 30% of DL vs. 3.2% of ulcers (OR = 12.9; 95%CI, 1.4-291; p0.01). Endoscopic treatment was required in 97% of DL and 39% of ulcers (OR = 47.5; 95%CI, 5.5-106.1; p0.0001). Active bleeding during endoscopy was registered in 81% of DL and in 29% of ulcers (OR = 10.2; 95%CI, 2.7-40.3; p0.0005). The remaining variables analyzed showed no significant differences. The mortality rate was 9.7% in the DL group and 3.2% in the ulcer group (p = NS). Multivariate analysis showed that gastric location (p0.01), active bleeding (p0.01), and comorbidity (p0.02) were more frequent in DL.Active bleeding and gastric location of the lesion were more frequent in the DL group than in the ulcer group. Diagnosis of DL is difficult, requiring a greater number of gastroscopies. Initial hemorrhage severity and the success rate of endoscopic treatment were similar in the two groups. The higher mortality found in DL can be explained by the greater comorbidity in these patients, with a worse bleeding tolerance.