Objective To compare the efficacy of S-adenosyl-L-methionine and ursodeoxycholic acid in improving serum biochemical abnormalities in gestational cholestasis. Design Randomised clinical trial.Setting University hospital.Population All women at <36 weeks of gestation with severe gestational cholestasis during June 1996 to December 2001. Methods Enrolled women were randomly assigned oral S-adenosyl-L-methionine 500 mg twice daily or oral ursodeoxycholic acid 300 mg twice daily until delivery. Main outcome measures Reduction in the concentration of serum bile acids. Other variables considered included obstetric and neonatal outcome, clinical symptoms and other laboratory measurements (serum levels of transaminases and bilirubin). The two groups were compared using Student's t test, Wilcoxon's signed rank sum test and Fisher's exact test, with a two-tailed P < 0.05 being considered significant. Results Of the 46 women enrolled, 24 received ursodeoxycholic acid and 22 S-adenosyl-L-methionine. At enrolment, gestational age, duration of therapy, rate of nulliparity, pruritus score and biochemical characteristics were similar between the groups. Both therapies significantly and equally improved pruritus. Women receiving ursodeoxycholic acid had a significantly greater improvement in the concentration of serum bile acids ( P ¼ 0.001), aspartate aminotransferase ( P ¼ 0.01), alanine aminotransferase (<0.001) and bilirubin ( P ¼ 0.002) compared with those receiving S-adenosyl-L-methionine. Duration of therapy was significantly greater in women receiving ursodeoxycholic acid compared with S-adenosyl-Lmethionine ( P ¼ 0.04), whereas gestational age at delivery and rate of prematurity were similar in the two groups. Conclusions In women with intrahepatic cholestasis of pregnancy, ursodeoxycholic acid is more effective than S-adenosyl-L-methionine at improving the concentration of serum bile acids and other tests of liver function, whereas both therapies are equally effective at improving pruritus.