Coronary heart disease (CHD) has a genetic component independent of traditional risk factors. However, the risk associated with individual genetic polymorphisms is likely modest, and therefore initial claims of associations with CHD require testing in other studies. We surveyed the literature for studies that retested polymorphisms previously reported to be associated with CHD, and found that these associations were reproduced for only 15 of the 172 polymorphisms considered. This review discusses how the modest risk associated with each individual polymorphism generates false-positive, as well as false-negative results. Common strategies to address these issues are described, including methods to aggregate risk estimates from several polymorphisms into a genetic risk score.