We have previously shown in a 2-yr controlled trial that hypercholesterolemia, frequent in primary biliary cirrhosis, is lowered by ursodeoxycholic acid (13 to 15 mg daily). To further investigate this effect, we analyzed the influence of long-term ursodeoxycholic acid administration on serum lipids, lipoproteins and bile acids. The study involved a subgroup of 33 noncirrhotic patients (17 received ursodeoxycholic acid and 16 received a placebo) analyzed at inclusion and after 2 yr. The total serum cholesterol concentration was markedly reduced in the ursodeoxycholic acid-treated patients in comparison with the controls (mean +/- S.E.M. = 7.49 +/- 0.42 mmol/L and 7.07 +/- 0.23 mmol/L at entry and 4.44 +/- 0.40 mmol/L and 6.89 +/- 0.27 mmol/L at 2 yr in the ursodeoxycholic acid and placebo groups, respectively; p < 0.02). Quantitatively, this decrease was mainly caused by a fall in low-density-lipoprotein cholesterol, but very low density-lipoprotein cholesterol levels also fell significantly. High-density-lipoprotein cholesterol levels remained stable in both groups, but the high-density-lipoprotein2/high-density-lipoprotein3 cholesterol ratio fell significantly during ursodeoxycholic acid treatment. No significant change occurred in total triglyceride or total phospholipid levels. In the treated group, the proportion of ursodeoxycholic acid increased (up to 60% of total circulating bile acids), whereas that of cholic and chenodeoxycholic acids fell significantly. In conclusion, the cholesterol-lowering effect of ursodeoxycholic acid could be related to an improvement of cholestasis, modifications in cholesterol metabolism or both. Changes in endogenous bile acid composition induced by ursodeoxycholic acid might be the common denominator of these two mechanisms.