Up to a third of patients have early rebleeding from oesophageal varices after endoscopic variceal ligation. Octreotide infusion is effective for control of variceal bleeding. We investigated the efficacy of octreotide infusion as an adjunct to endoscopic variceal ligation to prevent early rebleeding from varices. 100 consecutive patients admitted with endoscopically confirmed oesophageal varices and active bleeding or signs of recent haemorrhage were randomly assigned endoscopic variceal ligation alone or octreotide (50 micrograms intravenous bolus injection followed by intravenous infusion at 50 micrograms per h for 5 days) plus endoscopic variceal ligation. Three patients in each group were excluded. Bleeding was controlled in 44 of 47 patients who received variceal ligation alone and in 45 of 47 who received combined treatment (p = 1.0). Recurrent bleeding was documented in 18 (38% [24-52]) patients who received variceal ligation alone and in four (9% [3-21] who received combined treatment (p = 0.0007). The relative risk of rebleeding was lower (0.22 [0.08-0.60]) in the combined therapy group. Ten patients in the variceal ligation group and one in the combined therapy group required balloon tamponade for massive haematesis and haemodynamic instability (p = 0.0039). The in-hospital and 30-day mortality rates were higher in the variceal ligation group than in the combined therapy group (19 vs 9% and 23 vs 11%), but the differences did not reach significance. The relative risks of in-hospital (0.5 [0.04=5.3]) and 30-day (0.45 [0.17-1.2]) mortality were lower in the combined therapy group. Octreotide significantly reduces recurrent bleeding and the need for balloon tamponade in patients with variceal haemorrhage treated by endoscopic variceal ligation.