BACKGROUND: Studies of different vaccine constructs have demonstrated variable efficacy against Mycobacterium tuberculosis in animal models. Despite the fact that these vaccines have used one or another of a very small number of immunodominant antigens, a direct comparison of the relative efficacy of the antigens and delivery systems has been difficult, because the studies have used different parameters for assessment. METHODS: We compared the efficacies of the most commonly used vaccine constructs--adjuvanted protein, plasmid DNA, and live bacterial vectors--bearing the immunodominant secreted antigens early secreted antigen target-6 and antigen 85B, either alone or as a fusion protein. Mice were vaccinated with these constructs, and the effects of different delivery systems on protective efficacy (as assessed by survival studies and by monitoring bacterial load) and antigen-specific responses (including the contribution of CD4 and CD8 T cells to these responses) were assayed by various methods. RESULTS: The relative efficacy of different vaccines is dependent on the delivery system, the antigen, and the animal model. Likewise, the relative immunodominance of individual antigens in the fusion molecule is altered by the choice of delivery system. CONCLUSION: These results clearly demonstrate the importance of assessing vaccine function by use of multiple parameters and indicate which parameters are most reliable for assessing vaccine efficacy.