To evaluate prophylactic surgery for esophageal varices, a prospective randomized controlled trial was begun in 1980 by the Japanese Research Society for Portal Hypertension. Methods of operation included selective shunts and nonshunting interruption procedures. One hundred and twelve Japanese patients, in whom endoscopic findings suggested risk of bleeding but who had no bleeding episode, were randomly allocated to the operated group of 60 patients or nonoperated group of 52 patients. Nine patients with idiopathic portal hypertension, histologically proven noncirrhotic disease, which all fell in the operated group, were excluded from the study and the remaining 103 patients (51 operated and 52 nonoperated) were analyzed. Long-term follow-up of patients for a median of 49 mo with a maximum of 73 mo showed a total of 11 (22%) deaths, including 2 operative deaths, in the operated group compared with 23 (49%) deaths in the nonoperated group. The cumulative survival rate at 5 yr in the operated group was 72%, which was significantly higher than the 45% of the nonoperated group (p less than 0.05). The cumulative variceal bleeding rate at 5 yr was 7% in the operated group, which was significantly lower than that of the nonoperated group of 46% (p less than 0.001). It was concluded that portal nondecompression surgery was effective in preventing the variceal bleeding and in improving survival.