The incidence of coronary heart disease (CHD) shows a North to South gradient in Europe. We tested whether that gradient could be accounted for by the distribution of putative susceptibility genotypes. We correlated the published frequencies of susceptibility genotypes for the genes most often associated with CHD (ACE, AGT, APOE, F2, F5, MTHFR, PON1, and SERPINE1) with the incidence of the disease, controlling for the effects of smoking, systolic pressure, total cholesterol, and body-mass index. In three polymorphisms a negative correlation between the incidence of CHD and the frequency of a suceptibility genotype was observed. For ACE this correlation was significantly negative even when discounting classical susceptibility factors. This suggests that some alleles described as susceptibility factors cannot account for disease incidence at the population level. A genetic component must be added to the "Mediterranean paradox": genetic variants deemed to be risk factors for CHD show a geographical pattern uncorrelated with the disease incidence. This pattern can be understood from the history of populations which has shaped the genetic diversity of the European populations in North-South clines, similar to what is observed for CHD incidence, which will tend to create spurious correlations with polymorphisms related, or not related, to the disease.