The present trial compared the effectiveness and complications of intravenous somatostatin and vasopressin in treatment of variceal bleeding. Sixty-one cirrhotic patients with endoscopically proven active variceal bleeding were included. Both drugs were given as continuous intravenous infusions for 48 hr. Thirty patients received somatostatin (250 micrograms per hr after a bolus of 50 micrograms) and 31 vasopressin (0.4 units per min). Initial control of bleeding was achieved in 26 (87%) patients receiving somatostatin and in 23 (74%) of those treated with vasopressin. However, 10 patients [not significant statistically] in the somatostatin group and 5 in the vasopressin group rebled during treatment, after a mean of 15 and 20 hr, respectively. Therefore, complete control of bleeding during the 48 hr of therapy was achieved in 16 (53%) patients treated with somatostatin and in 18 (58%) of those receiving vasopressin. Mortality during hospitalization was similar in both groups (somatostatin 47%, vasopressin 45%). Differences were observed in complications associated with each therapy. Vasopressin produced major complications in 8 patients (left ventricular failure in 4 and severe abdominal pain requiring drug withdrawal in 4), and minor complications in 14; somatostatin infusion produced minor complications in 3 patients (p less than 0.01). In addition, the serum sodium concentration was significantly reduced by vasopressin (from 134.3 +/- 1.6 to 128.3 +/- 1.4 mEq per liter, p less than 0.001) but not by somatostatin (134.6 +/- 1.1 vs. 133.2 +/- 1.1 mEq per liter). This study shows that somatostatin is as effective as vasopressin in controlling variceal hemorrhage, but has a much lower rate of complications.