The major goal of the present study was to develop a limiting dilution system for the enumeration of T cells which respond to mycobacterial antigens. Purified T cells from M. tuberculosis-immune mice were restimulated with mycobacterial antigens and accessory cells, and after 4 days expanded with antigen, accessory cells and T cell growth factor. After another 3 days, proliferative responses were determined. Similar cultures performed without antigen served as controls. Limiting dilution analysis revealed that approximately 1/2000 to 1/3000 T cells from M. tuberculosis-immune mice responded to whole M. tuberculosis organisms while T cells from normal mice did not respond. Similar T cell numbers reacted with several mycobacterial strains indicating expression of shared T cell antigens. Using a semi-purified recombinant 64-kDa protein from M. bovis the frequency of T cells generated after immunization with M. tuberculosis which reacted with a single mycobacterial protein could be estimated. We found that approximately 1/5 of the M. tuberculosis-reactive T cells recognized this particular antigen. Immunization with the recombinant 64-kDa protein in an adjuvant containing trehalose dimycolate, monophosphoryl lipid A and mycobacterial cell wall skeleton stimulated an equally high number of M. tuberculosis-reactive T cells (1/2000). These findings demonstrate that a high proportion of tuberculosis-responsive T cells are directed against the 64-kDa protein and that immunization with this antigen in an appropriate adjuvant system is capable of stimulating high numbers of M. tuberculosis-reactive T cells. Limiting dilution analysis with a panel of mycobacterial proteins or peptides may allow their ranking from immunodominant to immunosilent and facilitate identification of antigens or epitopes relevant to protection.