In order to determine the effect of ursodeoxycholic acid on nonalcoholic fatty liver disease, 30 patients with body mass indices higher than 25, serum levels of alanine aminotransferase (ALT), aspartate aminotransferase (AST) or γ-glutamyltransferase (γ-GT) at least more than 1.5 times the upper limit of normality, and hepatic steatosis demonstrated by ultrasonography were randomized into two groups of 15 patients to receive placebo or 10 mg kg -1 day -1 ursodeoxycholic acid for three months. Abdominal computed tomography was performed to quantify hepatic fat content, which was significantly correlated with histological grading of steatosis (r s = -0.83, P < 0.01). Patient body mass index remained stable for both groups throughout the study, but a significant reduction in mean (± SEM) serum levels of ALT, AST and γ-GT was observed only in the treated group (ALT = 81.2 ± 9.7, 44.8 ± 7.7, 48.1 ± 7.7 and 52.2 ± 6.3 IU/l at the beginning and after the first, second and third months, respectively, N = 14, P < 0.05). For the placebo group ALT values were 66.4 ± 9.8, 54.5 ± 7, 60 ± 7.6 and 43.7 ± 5 IU/l, respectively. No alterations in hepatic lipid content were observed in these patients by computed tomography examination (50.2 ± 4.2 Hounsfield units (HU) at the beginning versus 51.1 ± 4.1 HU at the third month). These results show that ursodeoxycholic acid is able to reduce serum levels of hepatic enzymes in patients with nonalcoholic fatty liver disease, but this effect is not related to modifications in liver fat content.