A more effective vaccine will be essential if the global tuberculosis (TB) pandemic is ever to be controlled. A large number of new tuberculosis vaccines have been developed, representing the whole range of modern strategies for vaccine formulation and delivery. There is currently no alternative to testing these new vaccines in experimental animals challenged with virulent Mycobacteriurn tuberculosis in order to assess their protective efficacy. Although such testing is being carried out in several animal species (mice, guinea pigs, rabbits), all rational models include pulmonary challenge with a low dose of virulent mycobacteria. The quantitative measures for TB vaccines include increased survival, amelioration of clinical signs and symptoms (e.g. prevention of weight loss), decreased lesion size, reduction in bacillary loads in the lungs, and prevention of extrapulmonary dissemination and hematogenous reseeding of the lung. Although the ultimate objective of vaccination in humans is to prevent transmission to susceptible contacts, no such measurement is being used in animal studies of new vaccines. The validation of an immunological "correlate of protection" is urgently needed. Candidates for such a correlate include antigen-specific interferon-gamma production by T cells of the memory phenotype (CD45RB(high) or mycobacterial killing by macrophages co-cultured with immune T cells. Additional animal models must be developed for vaccines designed to prevent endogenous reactivation or exogenous reinfection, or to be used as a adjunct to chemotherapy.