We conducted a prospective, randomized trial comparing sclerotherapy and ligation in 120 patients with acute bleeding of esophageal varices. All the patients were cirrhotic, 59 received sclerotherapy, and 61 received ligation. Treatment was repeated regularly until the varices were obliterated. The mean follow-up period was 295 +/- 120 days and 310 +/- 105 days for the sclerotherapy and ligation groups, respectively. The control of active bleeding was 12/15 (80%) in the sclerotherapy group and 18/19 (94%) in the ligation group (P = .23). The numbers of treatment sessions required to achieve variceal obliteration were 6.5 +/- 1.2 in the sclerotherapy group and 3.8 +/- 0.4 in the ligation group (P < .001). Recurrent bleeding from the gastrointestinal tract was 51% in the sclerotherapy group compared with 33% in the ligation group (P < .05). Recurrent bleeding from esophageal varices was 36% in the sclerotherapy group and 11% in the ligation group (P < .01). However, bleeding from ectopic varices and congestive gastropathy was less common in the sclerotherapy group (7%) than in the ligation group (18%) (P = .05). Significant complications were encountered in 19% of the sclerotherapy group and in 3.3% of the ligation group (P < .01). Comparison of Kaplan-Meier estimates of time to death of both groups showed a significantly lower mortality in the ligation group (P = .011). Both sclerotherapy and ligation can effectively arrest active bleeding from esophageal varices. However, ligation is more effective than sclerotherapy in decreasing the risk of rebleeding from esophageal varices with fewer complications. Ligation can also achieve obliteration of esophageal varices more rapidly than sclerotherapy.(ABSTRACT TRUNCATED AT 250 WORDS)