Direct or ex vivo BMP9 adenoviral gene therapy can induce massive bone formation at the injection sites and clearly promote spinal fusion. A comprehensive analysis of the osteogenic activity indicated that BMP9 was one of the most potent inducers of osteogenic differentiation both in vitro and in vivo among 14 types of human BMPs. However, genetic variations and whether they correlated with OPLL were not considered. We have sequenced the complete BMP9 gene in 450 patients with OPLL and in 550 matched controls. Analyses were performed on single markers and haplotypes. Single marker tests identified 6 SNPs, among which the minor alleles of rs7923671 (T > C; P = 0.0026; OR: 1.33, CI: 1.10-1.60), rs75024165 (C > T, Thr304Met; P < 0.001; OR: 1.76, CI: 1.47-2.12) and rs34379100 (A > C; P < 0.001; OR: 1.52, CI: 1.27-1.82) were associated with OPLL. Logistic regression analysis showed that the additive model of rs75024165 (TT vs. CT vs. CC; P < 0.001; OR: 1.74) and rs34379100 (CC vs. AC vs. AA; P = 0.003; OR: 1.95) retained statistical significance when adjusted for clinical and demographic characteristics. Linkage disequilibrium (LD) analysis identified one 3 kb block of intense LD in BMP9 and one specific haplotype, CTCA (P < 0.001; OR: 2.37), that contained the OPLL-associated risk alleles and was a risk factor for OPLL. This haplotype is associated with increased severity of OPLL, as shown by the distribution of ossified vertebrae in patients with OPLL (P = 0.001). In summary, in the Chinese population studied, SNPs in the BMP9 gene appear to contribute to the risk of OPLL in association with certain clinical and demographic characteristics. The severity of OPLL seems to be mediated predominantly by genetic variations in a 3kb BMP9 locus with the specific haplotype CTCA.