Abstract: Objective To provide a scientific basis of tuberculosis (TB) prevention and control by analyzing the characteristics of single nucleotide polymorphism (SNP) of the whole-genome sequences of Mycobacterium tuberculosis. Methods The whole-genome sequences of 2 372 M. tuberculosis strains were download form National Center for Biotechnology Information (NCBI) and European Nucleotide Archive (ENA), we qualified the raw data to clean redundancy as specific rules required, BWA v 0.7.12, SAMtools v 1.3, Picard v 1.112 and Varscan were employed to call SNPs and mappability values were used to filter out non-unique SNP sites. The maximum likelihood analysis constructed the phylogenetic tree by software RAxMLv 8.2.8, and calculated the value of Fst for each SNP with software Genepop v 4.5.1. Finally, software SnpEff v 4.3c was used to annotate each SNPs site. Results We gained 107 654 sites by initially SNP calling, the phylogenetic tree based on the 107 654 SNP sites classified 2 347 isolates into 7 lineages and 69 sublineages. And we gained 285 SNP sites by optimizing, and the phylogenetic tree based on the 285 sites classified 2 347 isolates into 7 lineages and 67 sublineages. Conclusion In the study we found a set of phylogeny-based SNP sites, the 285 SNP sites can not only be used in the phylogenetic and evolutionary analyses, but also used as markers of genotyping in molecular epidemiological research of tuberculosis.