This study aimed to develop an epirubicin dose modification scheme in women with breast cancer and liver dysfunction. We first identified target areas under the concentration-time curve (AUCs) of 2400 and 1600 ng/ml.h from pharmacokinetic studies in 15 women with normal liver tests. In a second group of 16 women with abnormal liver biochemistry, the relationship between raised asparate aminotransferase (AST) and epirubicin clearance was: dose=AUC (97.5-34.2xlog AST). Adaptive dosing was evaluated prospectively in a third group of 41 women with serum AST > or =2xnormal+/-raised bilirubin. The median AUCs were 2444 and 1608 ng/ml.h, close to the high and low target AUCs, respectively. Variability in AUC was lower with adaptive dosing than in a fourth group given an unadjusted dose of epirubicin (coefficient of variation=25.8, 30.0 and 46.5%, respectively; P=0.06). Epirubicin dosing based on AST is safe and may reduce pharmacokinetic variability.