Numerous benefits for patients have been predicted if prescribing decisions were routinely accompanied by pharmacogenetic testing. So far, little attention has been paid to the possibility that the routine application of this new technology could result in considerable harm to patients. This article emphasises that pharmacogenetic testing shares both the opportunities and the pitfalls with 'conventional' disease-genetic testing. It demonstrates that performing pharmacogenetic tests as well as interpreting the results are extraordinarily complex issues requiring a high level of expertise. It further argues that pharmacogenetic testing can have a huge impact on clinical decisions and may influence the therapeutic strategy as well as the clinical monitoring of a patient. This view challenges the predominant paradigm that pharmacogenetic testing will predict patients' responses to medicines, but that it will not provide any other significant disease-specific predictive information about the patient or family members. The article also questions published proposals to reduce the consent procedure for pharmacogenetic testing to a simple statement that the physician wishes to test a sample of the patient's DNA to see if a drug will be safe or whether it will work, and presents an alternative model that is better suited to protect patient's interests and to obtain meaningful informed consent. The paper concludes by outlining conditions for the application of pharmacogenetic testing in clinical practice in a way that can make full use of its potential benefits while minimising possible harm to patients and their families.