We have compared the effect of ursodeoxycholic acid with placebo on the clinical state, blood liver chemistries and serum and urinary bile acids in four patients with primary biliary cirrhosis. All parameters were evaluated monthly, and bile acid composition was measured by capillary gas-liquid chromatography. At the time of admission, all patients showed intense pruritus, and their serum alkaline phosphatase, AST and ALT levels were elevated 4.3, 2.7 and 2.3 times over control values. Serum bile acids were elevated almost 38-fold with 2.5 times more cholic acid than chenodeoxycholic acid. Urinary bile acid output was elevated 28 times the control values, and 36% were 1β-hydroxycholic acid, 1β-hydroxydeoxycholic acid and hyocholic acid (3α,6α,7α-trihydroxy-5β-cholanoic acid). Three months of placebo administration did not significantly affect the clinical or biochemical presentations, and the serum and urinary bile acid composition did not change. In contrast, ursodeoxycholic acid feeding (12 to 15 mg per kg per day) for 6 months abolished pruritus in two and lessened itching in two subjects and reduced serum alkaline phosphatase, AST and ALT levels by 21, 35 and 47%, respectively. The mean values for the total serum bile acid concentrations in these patients declined 26% from the pretreatment value, but the proportion of ursodeoxycholic acid increased from 3 to 40% of the total bile acids; thus, total fasting serum endogenous bile acid levels decreased almost 50%. Similar changes were noted in the urinary bile acids, in which ursodeoxycholic acid became the major bile acid, and approximately 18% were hydroxylated at C-1, C-6 and C-21. These results demonstrate that, compared with placebo, ursodeoxycholic acid significantly lowers serum and urinary bile acids in primary biliary cirrhosis and produces considerable clinical and biochemical improvement.