Molecular epidemiology has been initiated for the confirmation of transmission link among tuberculosis patients. IS6I 10 restriction fragment length polymorphisms (RFLP) technique has been used as an excellent tool to discriminate Mycobacterium tuberculosis isolates, especially in tuberculosis (TB) outbreak in the population. IS6110 RFLP has the most discriminatory power for the M. tuberculosis isolates with high copy number of IS6110 like Korean isolates. Spoligotyping using spacers of direct repeat is useful to distinguish Beijing strains which are found widely in Eastern Asia, from non-Beijing strains. It is known that Beijing strains are more virulent, apt to be drug resistant than non-Beijing strains. Strain typing techniques of Mycobacterium tuberculosis has lead to the development of phylogenetic classification. Variable number tandem repeat (VNTR) of M. tuberculosis is another good target for strain typing. The technique using VNIR is rising as an alternative tool to overcome disadvantages of IS6110 RFLP which is time consuming in the sense that it takes longer time to process from the culture positive bacilli, and has the intrinsic difficulties in objectification of the results. The combination of many VNTR loci enhances discriminatory power to become equal to that of IS6110 RFLP. On the other hand, the optimal VNTR combination differs from one country to another due to different dominant clade. Large sequence polymorphisms (LSP) and single nucleotide polymorphisms (SNP) are important tools for the classification of the phylogeny of M. tuberculosis complex. Many previous reports indicate that the depending upon the type of strains, the ways of transmission of disease, the way to get infected with disease and the development of drug resistance conditions are variable. Therefore, the molecular epidemiology of M. tuberculosis has become more important for tuberculosis control in the world. It will be possible to set up tuberculosis-tailored policy after the characterization of M. tuberculosis by molecular epidemiologically.