We evaluated the in vivo therapeutic activities of benzoxazinorifamycin KRM-1648, clarithromycin (CAM) and levofloxacin (LVFX) against Mycobacterium avium infection induced in mice. Mice infected intravenously with M. avium (1.4 x 10(7)) were given KRM-1648 (20 mg/kg), CAM (10 mg/kg), or LVFX (5 mg/kg) alone, or combination of KRM-1648 with diclofenac sodium (1.25 mg/kg) by gavage, once daily, five times per week, from day 1 for up to 8 weeks. The bacterial loads in the lungs and spleens were determined by counting colony forming units of the organisms in the tissue homogenates of the visceral organs using 7H11 agar plates. Both KRM-1648 and CAM caused significant levels of bacteriological response in mice treated with these drugs, while LVFX exerted no appreciable therapeutic effect. The therapeutic efficacies of test antimicrobials were in the order, KRM-1648 > CAM > > LVFX. The combined use of diclofenac sodium with KRM-1648 did not affect the expression of therapeutic activity of KRM-1648. This excludes the possibility that cyclooxygenase-dependent inflammatory reactions may be involved in the establishment of persistent bacterial growth of M. avium organisms at the sites of infection in mice. Furthermore, the present study showed that the parameters of in vitro antimicrobial activities of drugs such as MIC and MBC values are not useful in predicting their therapeutic outcome in M. avium-infected mice.