Pyrazinamide (PZA) is a unique antituberculosis drug because it is effective in vivo but not in mediums commonly used to culture tubercle bacilli. Consequently, it was employed to test the validity of an in vitro macrophage model of human tuberculosis for value as a correlate of clinical events. The drug was as active in the macrophage model as it was clinically, inhibiting virulent tubercle bacilli at concentrations at 20 micrograms/ml or higher. By contrast, it was ineffective in 7H9 bacteriologic culture medium, even at concentrations as high as 2,560 micrograms/ml. It could be either bacteriostatic or bactericidal against intramacrophage tubercle bacilli, depending on its concentration, the donor of the macrophages, and the length of exposure of the infected macrophages to the drug. The data presented suggest that the clinical effectiveness of PZA is determined by a complicated self-modulating sequence of interactions between it, tubercle bacilli, and host macrophages. Specific evidence was found that tubercle bacilli may replicate within human macrophages in non-acid-fast form, thus indicating that colony-forming unit counts are inherently more accurate than acid-fast bacilli counts in these experiments, and also suggesting that important changes in bacillary cell wall composition may occur among tubercle bacilli within infected human macrophages.