One hundred and nine patients presenting with severe haemorrhage from benign peptic ulcers were randomised to either endoscopic injection sclerotherapy using a combination of 1:100000 adrenaline and 5% ethanolamine or to conservative treatment. Only high risk patients with active bleeding or endoscopic stigmata of recent haemorrhage and accessible ulcers were considered. The two groups were well matched for age, shock, haemoglobin concentration, endoscopic findings, and consumption of non-steroidal anti-inflammatory drugs. The group treated endoscopically had a significantly reduced rebleeding rate (12.5% v 47%, p<0.001). Rebleeding was successfully treated in some patients by injection sclerotherapy, other patients underwent urgent surgery. While there was a tendency towards a lower operation rate and lower transfusion requirements in the treated group, this failed to achieve statistical significance. The use of injection sclerotherapy in the conservatively treated group after rebleeding undoubtedly reduced the number of surgical operations. Endoscopic injection sclerotherapy is effective in the prevention of rebleeding in these patients.Bleeding peptic ulcer can be treated by a range of endoscopic techniques including Nd-Yag laser photocoagulation,`2 the heater probe,34 bipolar,56 or monopolar78 electrodes. These modalities tend to be expensive, require highly trained assistance and technical back up, and are not widely used. Recent attention has focused upon injection sclerotherapy of the bleeding area using adrenaline alone,9 a sclerosant alone,'0`or