To address the effect of ursodeoxycholic acid therapy on development of esophageal varices in patients with primary biliary cirrhosis. We compared, as part of a prospective treatment trial, the risk of varices developing in patients with primary biliary cirrhosis who received ursodeoxycholic acid (13 to 15 mg/kg daily) versus those who received placebo for up to 4 years. Upper endoscopy was performed every 2 years or as indicated clinically. At the end of the 4-year period, all patients in the placebo group were offered ursodeoxycholic acid therapy. During follow-up, the risk of developing endoscopically confirmed varices was assessed. The 180 patients who entered the ursodeoxycholic acid trial were assessed for the presence or absence of varices by esophagogastroduodenoscopy; 139 patients had no varices, and 41 patients demonstrated varices on initial examination. At 4 years, the risk of newly developing endoscopically confirmed varices was 16% for the ursodeoxycholic acid-treated patients and 58% for the placebo-treated patients (p = 0.001). Thus, the use of ursodeoxycholic acid was associated with a significantly lower risk of developing varices in patients with primary biliary cirrhosis. In addition to biochemical improvement, delay in death, and prolongation of time to orthotopic liver transplantation, ursodeoxycholic acid has now been demonstrated to decrease the risk of esophageal varices developing in patients with primary biliary cirrhosis.