This study prospectively compares multipolar electrocoagulation and injection therapy in high-risk patients with bleeding ulcers. Patients were considered for entry if they had a bloody nasogastric aspirate, melena, or hematochezia and unstable vital signs, transfusion of greater than or equal to 2 U of blood in 12 hours, or a decrease in hematocrit of greater than or equal to 6% in 12 hours. Sixty patients with endoscopic evidence of an ulcer with active bleeding (n = 26) or a nonbleeding visible vessel (n = 34) were randomly assigned to receive multipolar electrocoagulation or injection with absolute ethanol. Hemostasis was achieved in 14 of 14 actively bleeding patients with multipolar electrocoagulation vs. 10 of 12 (83%) treated with injection. No significant differences were observed between electrocoagulation and injection therapy in any parameter assessed during the hospitalization: incidence of further bleeding (6% vs. 10%), units of blood transfused after treatment (1.8 +/- 0.6 vs. 1.3 +/- 0.4), incidence of surgery for bleeding (6% vs. 7%), length of hospital stay in days (5.8 +/- 0.9 vs. 7.2 +/- 2.5), cost of hospitalization (+7160 +/- +1630 vs. +8520 +/- +2960), or mortality rate (3% vs. 3%). Treatment induced bleeding in nonbleeding visible vessels in 35% of subjects in each group, but this was controlled with continued treatment in all patients. One delayed perforation occurred 9 days after multipolar electrocoagulation. Multipolar electrocoagulation and injection therapy are of comparable efficacy in the treatment of patients with clinical evidence of a major upper gastrointestinal bleed and endoscopic evidence of an ulcer with active bleeding or a nonbleeding visible vessel.