Ninety-six patients with good liver function (Child class A or B) and esophageal varices were randomly assigned to one of three groups given different treatments: endoscopic injection sclerotherapy (n = 32), esophageal transection (n = 32) or distal splenorenal shunt (n = 32). Five patients (5.2%) had to be excluded from this study because severe chronic pancreatitis made separation of the distal splenic vein from the pancreatic bed difficult. Esophageal transection was performed for these patients. No deaths occurred during the 30 days of treatment. The 5-yr cumulative bleeding rates were 0%, 5.9% and 12.9% in the endoscopic injection sclerotherapy, esophageal transection and distal splenorenal shunt groups, respectively (no statistical significance). In no case in the three groups did death occur because of variceal bleeding. Sixteen patients died, mainly because of underlying liver disease; four were in the endoscopic injection sclerotherapy group, five were in the esophageal transection group and seven were in the distal splenorenal shunt group. No statistically significant difference in survival rate among the three groups was found. These results show that endoscopic injection sclerotherapy is a satisfactory alternative to esophageal transection or distal splenorenal shunt for the clinical management of patients with esophageal varices.