We performed a prospective, randomized trial to assess the efficacy of endoscopic injection therapy with absolute ethanol in preventing recurrent bleeding in patients with nonbleeding visible vessels in gastric ulcers. During the period of 1990 to 1993, 62 patients who bled were found to have gastric ulcers with nonbleeding visible vessels; all of them were enrolled for this trial. The 62 patients were randomly divided into two groups, which were comparable at entry. In group I (33 patients), we performed endoscopic injection therapy with absolute ethanol. In group II (29 patients), we sprayed the ulcers with 0.1% epinephrine and thrombin. Endoscopic injection therapy with ethanol was performed at the second endoscopy in the patients in both groups who had recurrent bleeding. Among the 33 patients in group I, 4 patients (12.1%) rebled after the initial ethanol injection therapy, whereas 10 of 29 patients (34.5%) rebled in the control group (p < .05). No patients in group I required surgical intervention, and ultimate hemostasis was achieved in all 33 group I patients (100%), indicating that endoscopic ethanol injection therapy achieves ultimate hemostasis and prevents recurrent bleeding in patients with gastric ulcers and nonbleeding visible vessels.