Partially fueled by the human immunodeficiency virus (HIV) pandemic, tuberculosis is the second-leading cause of infectious diseases–associated mortality globally and the leading cause of death among those infected with HIV. In 2010, there were an estimated 8.8 million incident cases of tuberculosis and approximately 1.5 million deaths attributed to tuberculosis. Increased public heath efforts have been effective in decreasing the overall prevalence and mortality associated with this disease, but declines in incidence rates have been outpaced by global population growth, such that absolute incidence rates have not declined as fast as previously expected [1]. Commensurate with increased public recognition of the importance of tuberculosis to global health in the mid-1990s, biomedical research for tuberculosis has increased both in the United States and internationally. Although tools to conduct modern biomedical research involving tuberculosis have only been available for a little over a decade, this research has resulted in several successes, including promising new vaccines to protect against tuberculosis in the highest-risk populations, diagnostic candidates to improve the accuracy and speed of diagnosing all forms of tuberculosis, and new drug candidates with the potential for improved treatment and chemoprevention for drug-sensitive and drug-resistant tuberculosis. Despite these successes, many questions remain in all aspects of tuberculosis research, ranging from fundamental pathogenesis to programmatic implementation, and they are complicated by the complexity of the disease and its chronic nature. Compared with other infections, Mycobacterium tuberculosis infection does not elicit clinically relevant disease in the majority of subjects, and many who are infected carry asymptomatic, paucibacillary disease for decades. This complicates the definition and evaluation of early infection events and initial host responses and the characterization of infecting bacillary populations. The clinical definition of latent infection is limited to positive responses to injection of purified mycobacterial components (ie, the tuberculosis skin test) or to T-cell–recall responses measured through the production of interferon γ; unlike viral infections, latent tuberculous infections cannot be identified on the basis of stable antibody titers. Progression to active disease is not a clear unidirectional process and is prolonged, and many affected persons appear to be able to clear the infection, whereas others progress to active disease. The spectrum of active tuberculosis is complex. Although adult pulmonary tuberculosis is considered of greatest relevance for public health, extrapulmonary disease occurs in a significant portion of subjects, and pediatric manifestations of tuberculosis may be mistaken for other pulmonary infections and are difficult to diagnose. Treatment of active disease is complicated by the fact that the current standard of care involves combinations of drugs whose administration is sequenced on the basis of the clearance of mycobacteria from sputum and that, unlike most other antibacterial treatments, have to be administered for a minimum of 6 months to elicit a stable cure. Hypotheses exist that explain the need for such prolonged therapy, but they are difficult to test experimentally because initial treatment of tuberculosis eliminates bacteria from sputum and thus limits options for studying bacterial populations. Unlike many other infectious diseases, in tuberculosis, patients do not become immune to reinfection at a later time. Thus, a clear definition of the markers of immune protection is dependent on successful trials of new vaccines or has to be derived from data for persons who were infected but did not progress to active disease. However, both criteria are complicated by the limitations listed above. Comprehensive reviews of research priorities that have been defined on the basis of the unique nature of tuberculosis and the associated scientific/technical limitations of studying the disease have been addressed elsewhere [2–5], but several issues merit further exploration.