We randomly assigned 95 patients with large esophageal varices (Grade 3 or 4) who had not previously had upper gastrointestinal tract bleeding to two groups: 49 received intravariceal sclerotherapy, and 46 were followed as controls. Over a mean follow-up of 13 months there was no difference between the sclerotherapy group and the control group in mortality (24.4 percent) or any significant difference in average hospital stay per month (3.0 vs. 2.6 days). Sclerotherapy was associated with significantly more episodes of upper gastrointestinal bleeding (26 vs. 10 episodes, P less than 0.05); 75 percent of deaths in the sclerotherapy group were related to bleeding, as compared with 18 percent in the control group. An additional 54 patients with cirrhosis who did not qualify for the study were also followed--20 with small varices and 34 with none. Mortality was 20 and 15 percent, respectively; no deaths were due to bleeding. We conclude that prophylactic sclerotherapy does not provide clinical benefit to patients with large esophageal varices.