The development of lesions in pulmonary tuberculosis, the most common form of tuberculosis, is the result of the conffict between the invader, Mycobacterium tuberculosis, and the host. This is a standard for all infectious diseases. Tuberculosis, however, is the paradigm for diseases with multiplication of the responsible organism within macrophages and monocytes (2) and control of the infection by cell-mediated immunity (CMI)(5) orchestrated by T-cell-derived lymphokines and carried out by the effector cells, activated macrophages (31). In fact, the CMI generated in tuberculosis is so potent that on average, 90% of the immunocompetent humans infected with M. tuberculosis are able to contain the infection and avoid progression to clinical disease during their lifetimes (8, 33, 34). This potency of CMI may be demonstrated experimentally: guinea pigs previously infected with M. tuberculosis react against cutaneous reinfection with live bacilli by rapidly forming a necrotic skin lesion that subsequently resolves spontaneously (31). This reaction, now known as the “Koch phenomenon,” suggests that the immune reaction mounted by infected animals is also particularly potent. In leprosy, a related disease, the responsible organism, M. leprae, is an obligate intracellular organism. It ensures its intracellular multiplication by inducing a specific immune paralysis, best demonstrated in the case of lepromatous leprosy, in which patients fail to develop delayedtype hypersensitivity (DTH) to the organism. In tuberculosis, while there may be immune dysregulation, specific DTH to purified protein derivative and granulomatous inffammation are prominent hallmarks of the disease. This contrast between M. leprae and M. tuberculosis begs the critical question of how M. tuberculosis avoids killing as a result of CMI despite its intracellular location. It is not simply an academic issue because it determines where antituberculous agents must encounter and kill M. tuberculosis during preventive and curative chemotherapy. To address this issue it is appropriate to review the remarkable but largely ignored work done in the past by Opie and Aronson (25), Long (15), Canetti (1), Lurie (18), and Dannenberg and Rook (7) and to recall the main steps of the pathogenesis of tuberculosis in humans, with emphasis on the events that take place after the development of acquired CMI.