Mice infected i.v. with the virulent Erdman strain of Mycobacterium tuberculosis exhibited three distinct phases of infection within the spleen. These consisted of a primary phase, characterized by the progressive growth of the organism; a secondary phase, in which the viable organism was progressively eliminated; and a tertiary phase, characterized by a chronic or slowly recrudescing disease state. Passive transfer experiments, in which T cell-enriched spleen cells from immune donors were infused into T cell-deficient recipients and were measured for their capacity to adoptively protect these mice from challenge with M. tuberculosis, provided evidence that at least three separate populations of protective T cells were acquired in response to the infection within the time frame of the experiments. These populations of T cells could be distinguished in that they differed in their expression of the L3T4 and Lyt-2 cell surface molecules, in terms of their kinetic profiles of emergence and loss, and (c) in terms of their susceptibility to cyclophosphamide. The results may suggest that different populations of protective T cells can be generated at different times during the infection as various classes of antigens (for example, metabolic or structural antigens) become available for presentation by host macrophages. It is hypothesized, furthermore, that the kinetics of emergence and loss of these various populations may reflect switching in the mode of immunity being expressed, particularly during the chronic phase of the infection, from that of a state of active immunity to one of immunologic memory.