An in vitro model of mycobacterial growth arrest was developed using Mycobacterium bovis BCG. When an exponentially growing culture was transferred to an evacuated tube, growth continued; treatment with a source of nitric oxide (diethylenetriamine-nitric oxide adduct [DETA-NO] at 50 microM) halted growth immediately, and aeration restored growth. When the period of growth arrest exceeded 4 h, a time lag occurred before aeration could restore growth. The lag time was maximal (24 h) after 16 h of growth arrest. These time lags indicated that one transition period was required for cells to achieve full arrest of growth and another for them to recover fully from growth arrest. DETA-NO-induced growth arrest failed to protect from the lethal effects of anaerobic shock, which caused rapid lysis of both growing and growth-arrested cells. While growth arrest had little effect on the lethal action of rifampin, it eliminated isoniazid lethality. Growth arrest reduced but did not eliminate fluoroquinolone lethality. Two fluoroquinolones, moxifloxacin and gatifloxacin, were equally lethal to exponentially growing cells, but moxifloxacin was more active during growth arrest. This difference is attributed to the fluoroquinolone C-7 ring structure, the only difference between the compounds. Collectively these data characterize a new system for halting mycobacterial growth that may be useful for evaluating new antituberculosis agents.