Mycobacterium tuberculosis is a successful pathogen that overcomes numerous challenges presented by the immune system of the host. This bacterium usually establishes a chronic infection in the host where it may silently persist inside a granuloma until, a failure in host defenses, leads to manifestation of the disease. None of the conventional anti-tuberculosis drugs are able to target these persisting bacilli. Development of drugs against such persisting bacilli is a constant challenge since the physiology of these dormant bacteria is still not understood at the molecular level. Some evidence suggests that the in vivo environment encountered by the persisting bacteria is anoxic and nutritionally starved. Based on these assumptions, anaerobic and starved cultures are used as models to study the molecular basis of dormancy. This review outlines the problem of persistence of M. tuberculosis and the various in vitro models used to study mycobacterial latency. The basis of selecting the nutritional starvation model has been outlined here. Also, the choice of M. smegmatis as a model suitable for studying mycobacterial latency is discussed. Lastly, general issues related to oxidative stress and bacterial responses to it have been elaborated. We have also discussed general control of OxyR-mediated regulation and emphasized the processes which manifest in the absence of functional OxyR in the bacteria. Lastly, a new class of protein called Dps has been reviewed for its important role in protecting DNA under stress. IUBMB Life, 57: 149–159, 2005 [ABSTRACT FROM AUTHOR]