I shall propose a new working hypothesis regarding the immunopathogenesis of tuberculosis infections in the mouse and guinea-pig models of pulmonary disease which differs from the current dogma that the development of caseous and potentially liquefied lesions in the susceptible guinea-pig lung is due to excessive expression of delayed-type hypersensitivity. I intend to show that the mouse is resistant because T cells entering the granulomatous lesions create highly organized wedges into the epithelioid macrophage fields, thus potentially saturating the tissues with cytokines. Where local necrosis does occur, it is contained by an efficient fibrotic response. In the guinea-pig, however, T cells remain in a peripheral mantle and do not invade the lesion. As a result, central macrophages are not activated, allowing bacterial growth and destructive caseation. I will describe data from the mouse system which show that certain mouse strains (C57BL/6, C5BL/10) are highly resistant to aerosol challenge, whereas other strains (DBA/2, AKR, CBA) are less so. These latter strains initially control the infection, but are subject to reactivation disease 50-200 days later. The histological data so far tend to support the hypothesis that these strains only induce acquired immunity during the reactivation phase, and upon initial exposure rely mainly on innate immunity to control the infection. This may be an important factor underlying the onset of reactivation disease in humans.