Bedaquiline (BDQ) has shown efficacy in shortening treatment duration and enhancing treatment success rates for multidrug-resistant tuberculosis (MDR-TB), thereby prompting widespread adoption. However, resistance to BDQ has emerged. This study aimed to identify genetic characteristics associated with decreased susceptibility to BDQ, using a public database to aid in the detection of resistant strains. Seventy-one BDQ-resistant and 929 BDQ-susceptible isolates from the open-source CRyPTIC database were selected for analysis. Variant calling was conducted via the clockwork pipeline. Univariate logistic regression was performed for each gene mutation, followed by LASSO regression for further variant selection. Ultimately, a multiple linear regression model was developed using log2-transformed Minimum Inhibitory Concentration values as the dependent variable, with variant selection refined through stepwise regression based on the Akaike Information Criterion. Ten gene mutations were significantly associated with reduced BDQ susceptibility, including two key gene mutations: Rv0678_141_ins_1 and Rv1979c_D249E, with effect estimates of 1.76 (95 % CI: 0.67–2.84) and 1.69 (95 % CI: 0.22–3.17), respectively. Other implicated genes included Rv2699c_-84_del_1, hsaB_I179T, mmpL9_T241A, pncA_C14R, Rv0373c_G621S, Rv0893c_L27F, Rv1770_A4D, and Rv3428c_S327C. This study identified ten gene mutations linked to decreased susceptibility to BDQ, providing a reference for developing a comprehensive catalog of BDQ-resistant genes.