A controlled, randomised study was performed to evaluate the efficacy of treatment with heater probe in the prevention of rebleeding from peptic ulcer with a non-bleeding visible vessel. One hundred and one patients were randomised into two groups: patients to be treated by heater probe (n = 51) and controls without active treatment (n = 50). In the heater probe group rebleeding occurred in five patients (10%) v 13 (26%) in the control group (p = 0.03), with a comparative risk of 0.38 in favour of the heater probe group. The difference in proportions of successful treatment for each group was 16.2% in favour of the heater probe (95% CI = 2 to 31%). Haemorrhage directly related to heater probe treatment occurred in four patients. In three of them bleeding was easily controlled by further heater probe pulses. There were no other complications and no death in the heater probe group. One patient in the control group died of pulmonary embolism. No significant differences in the length of stay in hospital, blood transfusions, surgical rates, or death were found; the design of the study, however, precluded an adequate assessment of these variables, because the heater probe was an optional rescue treatment when high surgical risk patients rebled. These results suggest that the heater probe is an effective and safe procedure in the prevention of recurrent haemorrhage in peptic ulcer with a non-bleeding visible vessel.