The possible involvement of a new chemical entity in pharmacokinetic drug interactions is an important safety issue. Not all relevant drug combinations for evaluation of the interactive potential of a new drug can be examined. Therefore, experiments should be selected to provide information which is valid not only for the interaction investigated, but which can be extrapolated to other comedications. In this respect the typical approaches currently used, including interaction studies with high risk drugs and compounds frequently given as comedication, or studies involving standard inhibitors and standard substrates are unsatisfactory. A better approach is to characterize drugs according to their effects on the underlying pharmacokinetic processes. Indeed, recent progress in understanding drug interaction mechanisms and in the development and refinement of in vitro test systems enables in many cases experiments to be designed which predict the occurrence of drug interactions. This paper illustrates systematic investigational procedures based on mechanism of interaction. Interaction mechanisms involving drug absorption, distribution, metabolism, and/or excretion are briefly summarized. Detailed proposals are derived which allow identification of the possible role of a new drug in interaction mechanisms for which valid test systems are available. Emphasis is placed on the rational selection of experiments with optimal cost-effectiveness. In vitro methods are integrated in the schemes wherever possible. In addition, it is proposed that pharmacoepidemiological screening, starting in phase II of drug development, be used to identify those relevant drug interactions missed by the mechanism-based approach. As exemplified by several recently discovered interactions it should be possible, by implementation of the proposed procedure, to avoid most serious unexpected adverse effects due to pharmacokinetic drug interactions.