Objective To analyze the mutation characteristics of isoniazid (INH) resistance gene of Mycobacterium tuberculosis (MTB)in Guangxi region, and provide basis for molecular diagnosis of drug-resistant tuberculosis. Methods 122 INH-resistant strains and 530 susceptible strains were obtained from the MTB strain library collected from 30 tuberculosis drug resistance monitoring points in Guangxi region, and underwent whole genome sequencing. Results Among 652 strains of MTB complex group, 127 (19.48%) had INH resistance gene mutations, including katG (15.64%, n=102), fabG1 (1.69%, n=11), ahpC (1.07%, n=7), kasA (0.61%, n=4) and inhA (0.46%, n=3) gene mutations. The coincidence rate of INH resistance phenotype and gene mutation was 90.03%, and the coincidence rate of INH resistance detected by proportional method and gene mutation detected by gene sequencing was not high (Kappa=0.677). There are 19 types of mutations, with single locus mutation accounting for 96.85% and combined mutation accounting for 3.15%. The locus with the highest mutation rate was katG315 (71.65%). The proportion of base change in the form of AGC-ACC was the highest (12.13%). The proportion of mutations in katG, ahpC and kasA genes in INH-resistant strains was higher than that in susceptible strains (all P < 0.05). There was no significant difference in the proportion of mutations in inhA and fabG1 genes between INH-resistant strains and susceptible strains (all P>0.05). Beijing strains and non-Beijing strains had the highest mutation rates of katG315, which were 18.75% (81/432) and 4.55% (10/220) respectively. Difference in distribution of mutation loci of katG gene in two genotypes of strains was statistically significant(χ²=16.253, P=0.039). Conclusion The mutation of INH resistance gene in Guangxi region is mainly at katG315 locus, and the INH resistance phenotype and gene mutation are not consistent. Mutations in katG, ahpC and kasA genes are associated with INH phenotypic resistance. Beijing genotype is associated with the mutation of katG gene.