During stationary growth or in vitro conditions mimicking relevant aspects of latency, the HspX protein (Rv2031c) is specifically upregulated by Mycobacterium tuberculosis. In this study we compared T-cell responses against HspX and the secreted M. tuberculosis protein Ag85B (Rv1886c) in tuberculosis (TB) patients, tuberculin skin test-positive individuals, M. bovis BCG-vaccinated individuals, and healthy negative controls. Gamma interferon responses to HspX were significantly higher in M. tuberculosis-exposed individuals than in M. tuberculosis-unexposed BCG vaccinees. In contrast, no such differences were found with respect to T-cell responses against Ag85B. Therefore, BCG-based vaccines containing relevant fragments of HspX may induce improved responses against this TB latency antigen. To identify relevant major histocompatibility complex class I- and class II-restricted HspX-specific T-cell epitopes, we immunized HLA-A2/K(b) and HLA-DR3.Ab(0) transgenic (tg) mice with HspX. Two new T-cell epitopes were identified, p91-105 and p31-50, restricted via HLA-A*0201 and HLA-DRB1*0301, respectively. These epitopes were recognized by human T cells as well, underlining the relevance of HspX T-cell recognition both in vivo and in vitro. In line with the data in humans, BCG immunization of both tg strains did not lead to T-cell responses against HspX-derived epitopes, whereas nonlatency antigens were efficiently recognized. These data support the notion that BCG vaccination per se does not induce T-cell responses against the latency antigen, HspX. Thus, we suggest that subunit vaccines incorporating HspX and/or other latency antigens, as well as recombinant BCG strains expressing latency antigens need to be considered as new vaccines against TB.