In vitro and in vivo studies have indicated that the induction or inhibition of cytochrome P450 (CYP) is one of the major mechanisms for some clinically important pharmacokinetic herb-drug interactions. Thus, an attempt was made to predict pharmacokinetic herb-drug interactions using the pharmacokinetic principles that are used for predicting drug-drug interactions. The expected AUC ratio was mainly dependent on unbound herbal inhibitor concentration ([I]) and inhibition constant (Ki), hepatic fraction (fh), number of inhibitory herbal constituents (n) and metabolic pathway fraction in hepatic metabolism (fm). Herb-drug interactions would be with low risk if sigma(i=1)n [[Ii]/Ki(i)] is less than 0.1, medium risk if it is between 0.1 and 1.0, and high risk if it is greater than 1. For high clearance drugs, the change of fh x fm had minor influence on AUC ratio when sigma(i=1)n [[Ii]/Ki(i)] values were fixed. Similarly, fm did not affect the AUC ratio for low clearance drugs. It appeared likely to predict a herb-drug metabolic interaction when [I], Ki, fh, fm and n could be determined. However, many herb- and drug-related factors may cause difficulties with the prediction, and well-designed human studies are always necessary.