Abstract: Objective To examine the potential causal relevance of high-density lipoprotein cholesterol（HDL-C）for coronary heart disease（CHD）using cholesteryl ester transfer protein（CETP）gene-629C/A variant as an instrument in a Mendelian randomization meta-analysis. Methods Eligible prospective or retrospective studies about the CETP-629C/A variant association with CHD and lipid profile were identified by searching PubMed and EMBASE till December 2015. Data were assessed in duplicate and analyzed by STATA software. Results A total of 12 nested case-control studies were eligible for the genotype-disease association and 9for the genotype-phenotype association. Overall analyses indicated that the-629 Callele was nonsignificantly associated with 3%（OR=1.03,95%CI0.93-1.14,P=0.603）,12%（OR=1.12,95% CI 0.95-1.32,P=0.186）and 11%（OR=1.11,95% CI0.96-1.28,P=0.174）increased risk of CHD under the allelic,homozygous genotypic and dominant models,respectively. In subgroup analyses,the corresponding effect estimates were significant in Caucasians,for myocardial infarction,in prospective and large studies. Plasma HDL-C was significantly reduced in carriers of-629 CC genotype（weighted mean difference or WMD-3.48;P＜0.01）or-629Callele（WMD-3.01,P＜0.01）. No significant changes were identified for plasma triglycerides. In the Mendelian randomization analysis,the significant causal odds ratio for a 1,5,10 mg/dL（1 mg/dL=0.026 mmol/L）genetic decrease of plasma HDL-C was 1.11,1.67,2.79 in Caucasians,1.07,1.43,2.03 for myocardial infarction,1.07,1.41,1.99 in prospective studies,1.07,1.38,1.90 in large studies,respectively. Conclusion This meta-analysis collectively reinforces the concept that genetically decreased plasma HDL-C is likely to be a potential causal risk factor for CHD,and this effect is more evident in Caucasians and for myocardial infarction.