One hundred patients with bleeding oesophageal varices were randomized into two treatment groups after resuscitation. One group was managed by tamponade only (group 1); the other group (group 2) was treated by endoscopic injection of oesophageal varices. The patients in group 2 were further subdivided into 25 patients (group 2a), who had tamponade applied immediately after sclerotherapy, and 25 patients (group 2b), who had sclerotherapy without subsequent tamponade. Injection of varices controlled the acute bleeding episode more effectively than tamponade (74 per cent in group 2 v. 42 per cent in group 1). There was no significant difference in the overall mortality rate of the two groups, but group 2 had a significantly higher proportion of Child's grade C patients (38/50 v. 29/50 = 76 v. 58 per cent). If only Child's grade C patients are considered, 16 out of 29 (55 per cent) died in group 1, whereas only 12 out of 38 (32 per cent) died in group 2 (P less than 0.05). Tamponade applied after sclerotherapy had no demonstrable effect on the outcome of sclerotherapy. The long term follow-up of patients (maximum 4 years) showed that recurrence of bleeding was less in the sclerotherapy group (8.1 per cent) than in the tamponade only group (27.6 per cent; P less than 0.05).