The ability to identify individuals who are susceptible to adverse drug reactions (ADRs) has the potential to reduce the personal and population costs of drug-related morbidity. Some individuals may show an increased susceptibility to certain ADRs through genetic polymorphisms that alter their responses to various drugs. We wished to establish a methodology that would be acceptable to members of the general population and that would enable estimation of the risks that specific genetic factors confer on susceptibility to specific ADRs. Buccal swabs were selected as a minimally invasive method to obtain cells for DNA extraction. We wished to determine whether DNA of sufficient quantity and quality could be obtained to enable genotyping for two different polymorphic genes that code for enzymes that are widely involved in drug disposition. This article describes a small pilot study of methodology developed in the New Zealand Intensive Medicines Monitoring Programme (IMMP) to link prescription event monitoring (PEM) studies with pharmacogenetics. The methodology involves a nested case-control study design to investigate whether patients with genetic variants in P-glycoprotein (P-gp) and cytochrome P450 (CYP) 2C9 are more susceptible to psychiatric or visual disturbances following cyclo- oxygenase-2 inhibitor use (ADR signals identified in the IMMP database) than matched control patients taking the medication without experiencing any ADRs. It was concluded that the use of buccal swabs is acceptable to patients and provides DNA of sufficient quantity and quality for genotyping. Although no differences in the distribution of genotypes in the case and control populations were found in this small study, case-control studies investigating genetic risks for ADRs using drug cohorts from PEM studies are possible, and there are several areas where population-based studies of genetic risk factors for ADRs are needed. Examples are discussed where research in large populations is required urgently. These are: (i) genetic variations affecting P-gp function; (ii) variations affecting drugs metabolised by CYP2C9 and other polymorphic CYP enzymes; (iii) genetic variation in beta-adrenergic receptors and adverse outcomes from beta-adrenoceptor agonist therapy; and (iv) genetic variation in cardiac cell membrane potassium channels and their association with long QT syndromes and serious cardiac dysrhythmias. Such studies will help to identify factors that increase the risk of unwanted outcomes from drug therapy. They will also help to establish in what circumstances genotyping should be performed prior to commencing drug treatment and in tailoring drug treatment for individual patients.