BACKGROUND: Blood lipid concentrations are causally related to the risk of coronary heart disease (CHD). Various associations between CHD risk and genes that moderately affect plasma lipid levels have been described, but previous studies have typically involved too few 'cases' to assess these associations reliably. METHODS: The present study involves 4685 cases of myocardial infarction (MI) and 3460 unrelated controls without diagnosed cardiovascular disease. Six polymorphisms of four 'lipid-related' genes were genotyped. RESULTS: For the apolipoprotein E epsilon2/epsilon3/epsilon4 polymorphism, the average increase in the plasma ratio of apolipoprotein B to apolipoprotein A(1) (apoB/apoA(1) ratio) among controls was 0.082 (s.e. 0.007) per stepwise change from epsilon3/epsilon2 to epsilon3/epsilon3 to epsilon3/epsilon4 genotype (trend P < 0.0001). The case-control comparison yielded a risk ratio for MI of 1.16 (95% CI: 1.06, 1.27; P = 0.001) per stepwise change in these genotypes. But, this risk ratio was not as extreme as would have been expected from the corresponding differences in plasma apoB/apoA(1) ratio between genotypes. Hence, following adjustment for the measured level of the plasma apoB/apoA(1) ratio, the direction of the risk ratio per stepwise change reversed to 0.83 (95% CI: 0.74, 0.92; P < 0.001). Similarly, for the apolipoprotein B Asn4311Ser and Thr71Ile polymorphisms, genotypes associated with more adverse plasma apolipoprotein concentrations were associated with significantly lower risk of MI after adjustment for the apoB/apoA(1) ratio. The B2 allele of the cholesteryl ester transfer protein TaqIb polymorphism was associated with a significantly lower plasma apoB/apoA(1) ratio, but with no significant difference in the risk of MI. Finally, the lipoprotein lipase Asn291Ser and T4509C (PvuII) polymorphisms did not produce clear effects on either the plasma apoB/apoA(1) ratio or the risk of MI. CONCLUSIONS: It remains unresolved why some of these genetic factors that produce lifelong effects on plasma lipid concentrations have significantly less than the correspondingly expected effects on CHD rates in adult life.