Background and aims: Multiple observational studies and small-scale intervention studies suggest that high vitamin K intake is associated with improved markers for cardiovascular health. Circulating phylloquinone solely represents phylloquinone (vitamin K1) intake, while dephosphorylated uncarboxylated Matrix Gla Protein (dp-ucMGP) represents both phylloquinone and menaquinone (vitamin K2) intake. This study aims to investigate the causal relationship between genetically predicted vitamin K concentrations and the risk of CHD via a two-sample Mendelian Randomization approach. Design: We used data from three studies: the European Prospective Investigation into Cancer and Nutrition (EPIC)-CVD case-cohort study, CARDIOGRAMplusC4D and the UK Biobank, resulting in 103,097 CHD cases. Genetically predicted vitamin K concentrations were measured using SNPs related to circulating phylloquinone and dp-ucMGP. We calculated a genetic risk score (GRS) including four SNPs (rs2108622, rs2192574, rs4645543 and rs6862071) related to circulating phylloquinone levels from a genome wide association study. Rs4236 was used as an instrumental variable for dp-ucMGP. Inverse-variance weighted (IVW) analysis was used to obtain Risk Ratios (RRs) for the causal relationship between phylloquinone and dp-ucMGP concentrations and CHD risk. Results: Using the genetic score for circulating phylloquinone, we found that circulating phylloquinone was not causally related to CHD risk (RR 1.00 (95%-CI: 0.98; 1.04)). Lower genetically predicted dp-ucMGP concentration was associated with a lower CHD risk with a RR of 0.96 (95%-CI: 0.93; 0.99) for every 10 μg/L decrease in dp-ucMGP. Conclusions: This study did not confirm a causal relationship between circulating phylloquinone and lower CHD risk. However, lower dp-ucMGP levels may be causally related with a decreased CHD risk. This inconsistent result may reflect the influence of menaquinones in the association with CHD.