INTRODUCTION: It has been widely reported that the slow acetylator phenotype of N-acetyltransferase 2 (NAT2) is associated with the development of antituberculosis drug-induced hepatotoxicity (ATDH). The aim of this report was to evaluate the level of agreement and accuracy of two recently recommended markers, the two-single nucleotide polymorphisms (SNP) (C282T and T341C) and tagSNP of NAT2 (rs1495741) genotypes, to predict the seven-SNP-inferred NAT2 phenotype in Bolivian and Argentinian tuberculosis (TB)-patient populations. In addition, we analyzed the association of these markers with ATDH. METHODS: We examined 331 TB patients who had been treated with anti-TB drugs. TagSNP of NAT2 genotyping was determined using PCR-restriction fragment length polymorphisms. The seven SNPs of NAT2 were determined using sequencing. Concordance analysis was carried out using Kendall's tau-b coefficient (w) and the degree of agreement with Cohen's κ coefficient (κ). Receiver operating characteristic receiver operating characteristic curves were obtained to measure the specificity and sensitivity of the method. A binary logistic regression was performed to identify variables associated with the development of ATDH. RESULTS: Both predictors showed a remarkable concordance (>95.0%) and an almost perfect agreement (κ>0.945; P<0.0001) with the predicted acetylator profile. However, the sensitivity of the tagSNP genotypes to predict the NAT2 acetylator phenotype was lower in the Bolivian population (92.3%) compared with the Argentinian population (100.0%). CONCLUSION: A nearly perfect agreement between both predictors and the predicted acetylation profile was observed with very high levels of sensitivity (>97%) and specificity (>98.0%). Furthermore, and as expected, both the two-SNP (C282T, T341C) and tagSNP were found to be independent variables in predicting ATDH with the same strength as seven-SNP of NAT2.