Fifty-two patients with Forrest Ia or Ib bleeding ulcers were randomized to receive endoscopic injection therapy with either 1:10,000 epinephrine in water (Group I) or distilled water (Group II). Twenty-five out of 27 patients in group I, versus 22 out of 25 patients in group II, achieved initial hemostasis after endoscopic injection therapy (p > 0.05). Five patients who did not respond to local injection had bleeding controlled by heater probe thermocoagulation or surgical intervention. Three patients in each group developed rebleeding after initial hemostasis. Four of these patients had bleeding controlled by surgical intervention, while the other two died of underlying diseases. No change in systemic blood pressure, but a significant drop in the pulse rate were noted in both groups after injection therapy. Patients with shock at admission or ulcer size greater than 2 cm had a significantly higher rebleeding rate after initial hemostasis than patients with normal blood pressure and ulcers under 2 cm (p < 0.05). No serious complications were observed after injection therapy, and no significant difference in the amounts of solution required for successful hemostasis was noted between the two groups. We conclude that a local tamponade with distilled water is as effective and safe as diluted epinephrine solution for endoscopic injection therapy.