We undertook a multicenter randomized trial to compare the efficacy of terlipressin combined with transdermal nitroglycerin and that of octreotide in the emergency control of acute variceal hemorrhage in cirrhosis. Over 16 mo, 87 patients with endoscopically proved active bleeding from esophageal or cardiac varices were enrolled in five centers in France and randomly assigned to receive intravenous terlipressin (2 mg and then 1 mg/4 hr over 24 hr) and transdermal nitroglycerin (10 mg/12 hr over 24 hr) (group 1) or octreotide (continuous intravenous infusion of 25 micrograms/hr over 12 hr and then 100 micrograms at hr 12 and hr 18 subcutaneously) (group 2). Initial control of bleeding was assessed at the end of 12 hr of treatment on the basis of stability of blood pressure and hematocrit level with no further transfusion requirements. At 12 hr, bleeding was controlled in 59% (24 of 41) in group 1 and 78% (36/46) of group 2 patients (Fisher's exact test, p = 0.064). Mean transfusion requirements over this 12-hr period were significantly greater in group 1 (three blood units; range = 0 to 13) than in group 2 (one blood unit; range = 0 to 5) (p = 0.002). After the first 12 hr, 20% of patients (5 of 24) had repeat bleeding in group 1 compared with 27% (10 of 36) in group 2. During the first 48-hr period, five patients (12%) died in group 1, compared with 3 (6%) in group 2. Few side effects were noted in either group. However, in group 1 two patients experienced severe bradycardia; it resulted in death in one patient.(ABSTRACT TRUNCATED AT 250 WORDS)