This controlled trial was designed to evaluate the prophylactic effect of nadolol on gastrointestinal bleeding in cirrhotic patients with large oesophageal varices who had never bled. Nadolol or placebo was given randomly to two groups of 53 patients. The percentage of patients free of gastrointestinal bleeding 1 year after inclusion in the study was 83 +/- 6% (mean +/- S.D.) in the nadolol group and 80 +/- 6% in the placebo group. In the nadolol and placebo groups, 40 and 47 patients, respectively, were compliant, i.e., took nadolol or placebo continuously. The percentage of patients who were free of bleeding 1 year after inclusion was 97 +/- 3% in the subgroup of compliant nadolol patients. This percentage was significantly higher than that of patients who were free of bleeding in the placebo group (P less than 0.03) as well as in the subgroup of compliant placebo patients (77 +/- 6%; P less than 0.02). We concluded that, although there was no overall significant effect of nadolol on the risk of bleeding in cirrhotic patients in good condition with large oesophageal varices, this study suggests that nadolol reduced the risk of bleeding in compliant patients.