The current diagnostic methods for tuberculosis (TB) have several limitations. Although commercial serological tests based on antibody detection are available, their variable accuracies limit their roles in the clinic. The aim of this study was to discover the improved biomarkers for TB disease by investigating the serum profiles of IgG and IgM antibodies against nearly all Mycobacterium tuberculosis (MTB) antigens in 36 active TB patients and 18 healthy controls (HCs) using proteome microarrays. Our results revealed that multiple antigens could induce stronger serum IgG or IgM responses in TB patients compared to HCs, among them, Rv2026c and Rv2421c were further validated by ELISA with sera from 221 samples and showed the moderate performance in diagnosing TB by receiver operating characteristic analysis. Moreover, logistic regression analysis was performed to establish a combined panel that provided better sensitivity and specificity at 82.5% and 88.12%, respectively, than single antigens in the diagnosis of active TB. Furthermore, the antibody reactivity against Rv2026c and Rv2421c was correlated with clinical backgrounds. These results suggest that the combination of different antigens and classes of antibodies could provide promise and encouragement in developing an efficient serological test for the diagnosis of active TB.