The in vitro and in vivo activities of a new benzoxazinorifamycin, KRM-1648 (KRM), against Mycobacterium tuberculosis were studied. The MIC at which 50% of the isolates are inhibited (MIC50) and the MIC90 of KRM for 30 fresh isolates of M. tuberculosis measured by the BACTEC 460 TB System were 0.016 and 2 micrograms/ml, respectively. These values were much lower than those for rifampin (RMP), which were 4 and >128 micrograms/ml, respectively, and considerably lower than those for rifabutin (RBT), which were 0.125 and 8 micrograms/ml, respectively. A correlational analysis of the MICs of these drugs for the clinical isolates revealed the presence of cross-resistance of the organisms to KRM and either RMP or RBT although the MICs of KRM were distributed over a much lower range than were those of the other two drugs. KRM and RMP at concentrations of 1 to 10 micrograms/ml almost completely inhibited the bacterial growth of RMP-sensitive strains (H37Rv, Kurono, and Fujii) of M. tuberculosis phagocytosed in macrophage-derived J774.1 cells. KRM was more active than RMP in inhibiting the growth of the RMP-resistant (MIC = 8 micrograms/ml) Kurata strain but failed to show such an effect against the RMP-resistant (MIC >128 micrograms/ml) Watanabe stain. When KRM was given to M. tuberculosis-infected mice at dosages of 5 to 20 mg/kg of body weight by gavage, one daily six times per week from day 1 after infection, it was much more efficacious than RMP against infections induced in mice by the RMP-sensitive Kurono strain, as measured by a reduction of rates of mortality, a reduction of the frequency and extent of gross lung lesions, histopathological changes in lung tissues, and a decrease in the bacterial loads in the lungs and spleens of infected mice. KRM also displayed significant therapeutic efficacy against infection induced by the RMP-resistant Kurata strain, while neither KRM nor RMP was efficacious against infection by the RMP-resistant Watanabe strain. In the case of infection with the Kurono strain, the efficacy of the drugs in prolonging the time of survival was in the order KRM, RBT, RMP. KRM was much more efficacious than RMP, when given at 1- to 4-week intervals. These findings suggest that KRM may be useful for the clinical treatment of tuberculosis contracted through RMP-sensitive strains, even when it is administered at long intervals.