We describe a detailed study of the effects of ursodeoxycholic acid administration on bile acid composition of the serum and bile of patients with primary biliary cirrhosis. Gas chromatography-mass spectrometry was used to analyze bile acids from 10 patients with primary biliary cirrhosis before and during ursodeoxycholic acid administration (500 mg/day, corresponding to approximately 8 mg/kg body wt), after group separation of the unconjugated and conjugated fractions by lipophilic anion exchange chromatography. These studies were directed at assessing whether the beneficial role of ursodeoxycholic acid in primary biliary cirrhosis was the consequence of a shift in the hydrophobic/hydrophilic balance of the bile acid pool and whether the hypercholeresis might result from the cholehepatic circulation of unconjugated ursodeoxycholic acid in bile. In basal conditions, the unconjugated bile acids accounted for only 5.5% and 2.5%, respectively, of the total bile acids of serum and bile; cholic acid was the major component of the conjugated fraction of serum and bile (56.0% ± 4.0%, mean ± S.E.M.), and ursodeoxycholic acid was present in only trace amounts. The conjugated fraction contained many unusual bile acids (representing 16.5% ± 1.3% of total) including C 25 bile acids, iso-chenodeoxycholic acid and several oxo-bile acids. After ursodeoxycholic acid administration biochemical indices of liver function all improved, but the proportions of the unconjugated bile acids in serum and bile did not significantly change. Ursodeoxycholic acid became the predominant biliary bile acid of conjugated bile acid fraction (33.4% ± 1.4%) and significant decreases occurred (p < 0.05) in biliary levels of cholic acid (28.3% ± 2.3%) and the unusual bile acids (9.2% ± 0.8%), but no changes occurred in the proportions of chenodeoxycholic acid (from 19.4% ± 2.7% to 19.4% ± 1.3%) and deoxycholic acid (from 5.4% ± 1.4% to 6.8% ± 1.3%). The proportions of biliary lithocholic acid increased significantly after ursodeoxycholic acid administration (from 0.7% ± 0.1% to 2.9% ± 0.9%; p < 0.05). In serum, ursodeoxycholic acid concentrations attained levels of 64 μmol/L. Serum lithocholic acid concentrations increased, whereas cholic and chenodeoxycholic acids decreased. Relative percentages of individual biliary bile acids in the unconjugated fraction were similar to the conjugated fraction before and after ursodeoxycholic acid treatment. The proportions of biliary unconjugated ursodeoxycholic acid did not change significantly after its administration (from 0.6% to 0.9% of the total). These data provide little support for a significant shift in the hydrophilicity of the circulating bile acid pool after ursodeoxycholic acid administration. However, because the detergency and hepatotoxicity of a bile salt appears to correlate well with the retention index determined by reverse-phase liquid chromatography, the qualitative changes observed in the biliary pool (i.e., a displacement of cholic acid by ursodeoxycholic acid), suggests that a relative lowering of the hepatotoxicity index of the bile acid pool occurs after ursodeoxycholic acid therapy. Furthermore, our findings indirectly suggest that a hypercholeresis related to the biliary secretion of unconjugated ursodeoxycholic acid does not play a role in the beneficial effects of ursodeoxycholic acid therapy in primary biliary cirrhosis.