Background:Epidemiologically, high-density lipoprotein (HDL) cholesterol levels associate inversely with risk of ischemic cardiovascular disease. Whether this is a causal relation is unclear.Methods:We studied 10,281 participants in the Copenhagen City Heart Study (CCHS) and 50,523 participants in the Copenhagen General Population Study (CGPS), of which 991 and 1,693 participants, respectively, had developed myocardial infarction (MI) by August 2010. Participants in the CCHS were genotyped for all six variants identified by resequencing lecithin-cholesterol acyltransferase in 380 individuals. One variant, S208T (rs4986970, allele frequency 4%), associated with HDL cholesterol levels in both the CCHS and the CGPS was used to study causality of HDL cholesterol using instrumental variable analysis.Results:Epidemiologically, in the CCHS, a 13% (0.21 mmol/liter) decrease in plasma HDL cholesterol levels was associated with an 18% increase in risk of MI. S208T associated with a 13% (0.21 mmol/liter) decrease in HDL cholesterol levels but not with increased risk of MI or other ischemic end points. The causal odds ratio for MI for a 50% reduction in plasma HDL cholesterol due to S208T genotype in both studies combined was 0.49 (0.11-2.16), whereas the hazard ratio for MI for a 50% reduction in plasma HDL cholesterol in the CCHS was 2.11 (1.70-2.62) (P(comparison) = 0.03).Conclusion:Low plasma HDL cholesterol levels robustly associated with increased risk of MI but genetically decreased HDL cholesterol did not. This may suggest that low HDL cholesterol levels per se do not cause MI.