Previous studies in murine and human models have suggested an important role for HLA Ia-restricted CD8+ T cells in host defense to Mycobacterium tuberculosis (Mtb). Therefore, understanding the Ags presented via HLA-Ia will be important in understanding the host response to Mtb and in rational vaccine design. We have used monocyte-derived dendritic cells in a limiting dilution analysis to generate Mtb-specific CD8+ T cells. Two HLA-Ia-restricted CD8+ T cell clones derived by this method were selected for detailed analysis. One was HLA-B44 restricted, and the other was HLA-B14 restricted. Both were found to react with Mtb-infected, but not bacillus Calmette-Guérin-infected, targets. For both these clones, the Ag was identified as culture filtrate protein 10 (CFP10)/Mtb11, a 10.8-kDa protein not expressed by bacillus Calmette-Guérin. Both clones were inhibited by the anti-class I Ab and anti-HLA-B,C Abs. Using a panel of CFP10/Mtb11-derived 15-aa peptides overlapping by 11 aa, the region containing the epitopes for both clones has been defined. Minimal 10-aa epitopes were defined for both clones. CD8+ effector cells specific for these two epitopes are present at high frequency in the circulating pool. Moreover, the CD8+ T cell response to CFP10/Mtb11 can be largely accounted for by the two epitopes defined herein, suggesting that this is the immunodominant response for this purified protein derivative-positive donor. This study represents the first time CD8+ T cells generated against Mtb-infected APC have been used to elucidate an Mtb-specific CD8+ T cell Ag.