In a period of 14 months, a prospective, randomized, controlled trial was undertaken in 61 patients who had bled from peptic ulcers in which, at endoscopy, non-bleeding vessels were visible in the ulcer crater. The control group consisted of 31 patients in whom the procedure was limited to observation alone; in 30 patients the ulcer base was coagulated by means of a heat probe. There were no statistically significant differences between the two groups in the rate of rebleeding, the assurance of ultimate hemostasis, the volume of blood transfusion required, the duration of hospitalization, or eventual mortality. However, the need for emergency surgical intervention was less frequent in the heat probe group (2 of 30) than among the controls (9 of 31) (p = 0.0243). Moreover, in the treatment group, those patients of advanced age (greater than 60 years), in shock, requiring blood transfusion of greater than 500 ml, and in whom blood or coffee ground material was seen in the stomach, thermocoagulation achieved statistically more effective hemostasis than that observed in the control group. For these high-risk subsets of patients with non-bleeding visible vessels, we recommend heat probe coagulation rather than mere observation.