A prospective randomized controlled trial of endoscopic heater probe therapy in bleeding peptic ulcers was performed to determine whether probe therapy can reduce rebleeding rates. Of 630 patients endoscoped for suspected upper gastrointestinal haemorrhage over a 16-month period, 166 (26 per cent) were found to have a peptic ulcer. Either minor or no stigmata of recent haemorrhage were found in 115 patients at the time of endoscopy. A single peptic ulcer with either active haemorrhage or a visible vessel was found in 51 patients, 43 of whom were entered into the trial. There were eight exclusions: four were inaccessible, one was a torrential haemorrhage and three were excluded for non-technical reasons. Patients were randomized to receive either heater probe (n = 20) or sham (n = 23) therapy. In actively bleeding ulcers, immediate haemostasis was achieved following probe therapy in 14 of 18 patients (78 per cent) compared with none of 21 having sham treatment (P less than 0.002). No rebleeding occurred in the probe therapy group (n = 20) compared with rebleeding in five of 23 sham treated patients (P = 0.05). Urgent surgery for haemostasis was required in three of the five sham treated patients who rebled. It is concluded that heater probe therapy may be effective in reducing rebleeding rates in peptic ulcers accessible to the endoscope.