The pharmacokinetics and clinical activity of epirubicin were investigated in 16 patients with hepatocellular carcinoma (HCC) who received epirubicin at 75 mg/m2; the drug was given intravenously to 7 patients and via the hepatic artery to 9 patients (7 of whom also underwent embolisation). Lignocaine (1 mg/kg) was also given intravenously to 15 patients, and the metabolite monoethylglycinexylidide (MEGX) was measured as an indicator of liver function. Epirubicin clearance correlated with serum aspartate aminotransferase (AST), albumin and bilirubin values in patients treated intravenously or intraarterially. Although the route of administration did not affect the median total plasma clearance of epirubicin, early- and intermediate-phase clearance was higher following intraarterial administration. MEGX levels correlated with serum bilirubin levels but there was no correlation with albumin or AST values or epirubicin clearance. The rate of response to epirubicin was 3/13 (23%; 95% confidence interval, 8%-50%). Intravenous epirubicin was tolerated well, but intraarterial treatment was associated with significant morbidity. These data confirm that although current recommended dose adjustments are based primarily on serum bilirubin levels, altered epirubicin pharmacokinetics correlate more strongly with AST and albumin values than with serum bilirubin concentrations. However, at this dose and schedule, epirubicin has only modest activity against HCC.