Screening for new antimicrobial agents is routinely conducted only against actively replicating bacteria. However, it is now widely accepted that a physiological state of nonreplicating persistence (NRP) is responsible for antimicrobial tolerance in many bacterial infections. In tuberculosis, the key to shortening the 6-month regimen lies in targeting this NRP subpopulation. Therefore, a high-throughput, luminescence-based lowoxygen-recovery assay (LORA) was developed to screen antimicrobial agents against NRP Mycobacterium tuberculosis. M. tuberculosis H 37 Rv containing a plasmid with an acetamidase promoter driving a bacterial luciferase gene was adapted to low oxygen conditions by extended culture in a fermentor with a 0.5 headspace ratio. The MICs of 31 established antimicrobial agents were determined in microplate cultures maintained under anaerobic conditions for 10 days and, for comparative purposes, under aerobic conditions for 7 days. Cultures exposed to drugs under anaerobic conditions followed by 28 h of "recovery" under ambient oxygen produced a luminescent signal that was, for most compounds, proportional to the number of CFU determined prior to the recovery phase. No agents targeting the cell wall were active against NRP M. tuberculosis, whereas drugs hitting other cellular targets had a range of activities. The calculated Z factor was in the range of 0.58 to 0.84, indicating the suitability of the use of LORA for high-throughput assays. This LORA is sufficiently robust for use for primary high-throughput screening of compounds against NRP M. tuberculosis.It is now widely accepted that a physiological state of nonreplicating persistence (NRP) is responsible for antimicrobial tolerance in many bacterial infections (11). In tuberculosis, a subpopulation of Mycobacterium tuberculosis isolates in NRP is considered an important contributing factor to the long treatment duration required, and the key to shortening the currently recommended 6-month regimen (the primary goal of new anti-M. tuberculosis chemotherapy) lies in effective targeting of this phenotype (2, 6). Standard drug susceptibility assays for detection of the activities of drugs against rapidly growing bacteria may not identify such compounds (8,11). In vitro models of M. tuberculosis isolates in NRP exist (4,30,(37)(38)(39), and there are several reports of studies that have assessed drug activity against NRP or stationary-phase cells (20,(34)(35)(36)42); however, these have relied upon the enumeration of CFU, thus precluding high-throughput screening (HTS) applications and requiring a minimum of 3 to 4 weeks for the completion of testing. In addition, the culture and preparation of NRP M. tuberculosis cells (5,17,29,30,38) have used batch cultures, which are not optimal for monitoring and for comparative studies. In contrast, a chemostat or a fermentor with a continuous-oxygen-monitoring culture system allows bacteria to be grown in a controlled and defined environment; there are two reports of the successful cultivation of M. tuberculos...