OBJECTIVE: Pharmacogenomics, the individualization of drug therapy based on genetic information, promises to change the clinical use of therapeutics and influence the drug development process, however the potential clinical and economic outcomes of pharmacogenomics is not clear. Our objective is to develop a framework for evaluating the cost-effectiveness of genetic-guided drug therapy. METHODS: We reviewed examples of currently practiced non-genetically based individualization of drug therapy. These examples were classified into two categories: dose modification and drug selection. The costs, response characteristics, and implementation strategy for the tests were evaluated. We also evaluated the therapeutics and diseases for which these tests were used to determine which characteristics led to the implementation of individualized drug therapy. We then developed a framework for evaluating the cost-effectiveness of pharmacogenomics and applied it to examples of currently debated pharmacogenomic strategies. RESULTS: Drug monitoring is typically employed only for drugs with a narrow therapeutic index, whereas drug selection is often employed with the use of antimicrobials, where drug effectiveness is highly dependent on diagnostic information. The economic and, importantly, the time cost are important determinates of the usefulness of testing procedures. The prevalence of polymorphisms, the penetrance of disease-related genes, and the magnitude of relative risk will also have a significant impact on the usefulness of genetic diagnostics. We found that some commonly cited examples of potential applications of pharmacogenomics do not fulfill these criteria. Our analysis suggests that drug-gene interactions such as warfarin-CYP2C9, 6-mercaptopurine-TPMT, and COX-II inhibitors-Familial Adenomatous Polyposis (FAP) have the potential to produce cost-effective results. CONCLUSIONS: The use of pharmacogenomics to improve the cost-effectiveness of pharmaceuticals requires careful evaluation and will not be useful in all cases.