The value of sclerotherapy as prophylaxis against the first episode of variceal hemorrhage has not been established. Therefore, we randomly assigned 133 patients with cirrhosis of the liver (of alcoholic origin in 66 percent), esophageal varices, and no previous intestinal bleeding to either prophylactic sclerotherapy (n = 68) or no prophylaxis (n = 65). The groups were comparable in hepatic function, endoscopic findings, and the pathogenesis of cirrhosis. All patients who subsequently had a first episode of variceal hemorrhage received sclerotherapy whenever possible. During a median follow-up of 22 months, variceal hemorrhage occurred in 28 percent of the patients receiving sclerotherapy and 37 percent of the controls (P = 0.3). Thirty-five percent of the sclerotherapy group and 46 percent of the control group died. The survival curves (Kaplan-Meier) of both groups were similar (P = 0.2). However, among patients with alcoholic and moderately decompensated cirrhosis (Child-Pugh group B), survival was significantly higher in those receiving sclerotherapy, although the risk of bleeding was only marginally reduced by this procedure. We conclude that prophylactic sclerotherapy does not significantly reduce the risk of bleeding from esophageal varices, but that a subgroup of patients with esophageal varices and moderately decompensated alcoholic cirrhosis may benefit from prophylactic sclerotherapy because of factors not solely attributable to prevention of an initial episode of variceal bleeding.