Background-Brain-derived neurotrophic factor (BDNF) is a pleiotropic peptide involved in maintaining endothelial integrity. It is unknown if circulating BDNF levels are associated with risk of cardiovascular disease (CVD). Methods and Results-We prospectively investigated the association of circulating BDNF levels with cardiovascular events and mortality in 3687 participants (mean age 65 years, 2068 women) from the Framingham Heart Study (FHS). Using a common nonsynonomous single nucleotide polymorphism (SNP) in the BDNF gene (rs6265), we then performed a Mendelian randomization experiment in the CARDIoGRAM (Coronary ARtery DIsease Genome-Wide Replication And Meta-Analysis) consortium (> 22 000 coronary artery disease [CAD] cases, > 60 000 controls) to investigate whether SNP rs6265 was associated with CAD in CARDIoGRAM and, if so, whether the effect estimate differed from that predicted based on FHS data. On follow-up (median 8.9 years), 467 individuals (261 women) in FHS experienced a CVD event, and 835 (430 women) died. In multivariable-adjusted Cox regression, serum BDNF was associated inversely with CVD risk (hazard ratio [HR] per 1-SD increase 0.88, 95% CI 0.80 to 0.97, P=0.01) and with mortality (HR 0.87, 95% CI 0.80 to 0.93, P=0.0002). SNP rs6265 was associated with BDNF concentrations (0.772 ng/mL increase per minor allele copy) in FHS. In CARDIoGRAM, SNP rs6265 was associated with CAD (odds ratio 0.957, 95% CI 0.923 to 0.992), a magnitude consistent with the predicted effect (HR per minor allele copy 0.99, 95% CI 0.98 to 1.0; P=0.06 for difference between predicted and observed effect). Conclusion-Higher serum BDNF is associated with a decreased risk of CVD and mortality. Mendelian randomization suggests a causal protective role of BDNF in the pathogenesis of CVD.