The correlation between biliary and serum levels of ursodeoxycholic and chenodeoxycholic acids was studied in a double-blind controlled manner in 39 patients before and during treatment with ursodeoxycholic acid, 800 mg/day; ursodeoxycholic acid, 400 mg/day; chenodeoxycholic acid, 750 mg/day; chenodeoxycholic acid, 375 mg/day; and placebo, respectively. On a total of 74 occasions, fasting duodenal bile and venous blood samples were obtained simultaneously. Biliary bile acid composition was determined by gas-liquid chromatography and serum ursodeoxycholic and chenodeoxycholic acid concentrations by radioimmunoassays. There was a much closer correlation between the biliary and serum levels of ursodeoxycholic acid (r = 0.8184, P less than 0.001) than between those of chenodeoxycholic acid (r = 0.4707, P less than 0.01). In contrast to serum chenodeoxycholic, which showed many overlaps between pre- and posttreatment values, serum ursodeoxycholic acid proved to be a very sensitive, specific, and convenient means of predicting the presence of increased levels of ursodeoxycholic acid in the enterohepatic cycle.