Only 50-75% of patients are estimated to benefit from successful drugs. Pharmacogenetics research has demonstrated that the risk of therapeutic failure, drug toxicity and drug interactions is not the same for everyone and that genetic variation is an important determinant of these events. With the maturation of the Human Genome Project, a rational approach to new and better therapies is now viewed as a realistic prospect that will result in drugs personalized to small, genetically defined groups of patients or even to individuals. If treatment is guided by genetic profiles individualized to specific drugs, we expect to reduce the frequency and severity of adverse drug reactions. A large set of genetically polymorphic markers predictive of human drug response is now available. Foremost among these markers is the family of pharmacokinetic polymorphisms mainly consisting of the polymorphic human drug metabolizing enzymes. So far, clinical application of this new knowledge has been limited. In this commentary, we describe the Marshfield program in personalized medicine now under way in Wisconsin, and the United States Patient Safety Initiative as the first two attempts to apply pharmacogenetics to improve pharmaceutical intervention in human disease in the United States in large patient populations.