The strategy for using streptomycin against tuberculosis assumes that it is not effective intracellularly. But according to animal cell experiments, this is probably incorrect. We retested this assumption with a new experimental model using cultured human macrophages infected with tubercle bacilli so that the results would be directly relevant to human disease. At 5 and 50 micrograms/ml, streptomycin inhibited the bacilli strongly and killed some; at the lowest tested concentration of 0.5 micrograms/ml, it inhibited them weakly. It was acting intracellularly, because it could inhibit even when added 2 days after the macrophages had been infected and washed free of extracellular bacilli, and because in our experimental model the bacilli were shown to be unable to multiply extracellularly. However, as has been reported for animal macrophages, the antibiotic was quantitatively more than 2 orders of magnitude less effective in human macrophages than in simple bacteriologic medium. Probably this is because streptomycin is concentrated within lysosomes where low pH greatly inhibits it. The human macrophage-tubercle bacillus chemotherapeutic bioassay we describe here for the first time could be a superior patient-consonant new method for testing antituberculosis agents and treatment regimens. It retains important in vivo features, the complete host cell-parasite relationship for instance, without giving up the in vitro advantages of rapidity and objectivity.