Although a consensus has emerged that an HIV vaccine should elicit a cytotoxic T lymphocyte (CTL) response, the characteristics of an effective vaccine-induced T lymphocyte response remain unclear. We explored this issue in the simian human immunodeficiency virus/rhesus monkey model in the course of assessing the relative immunogenicity of vaccine regimens that included a cytokine-augmented plasmid DNA prime and a boost with DNA or recombinant pox vectors. Recombinant vaccinia virus, recombinant modified vaccinia Ankara (MVA), and recombinant fowlpox were comparable in their immunogenicity. Moreover, whereas the magnitude of the peak vaccine-elicited T lymphocyte responses in the recombinant pox virus-boosted monkeys was substantially greater than that seen in the monkeys immunized with plasmid DNA alone, the magnitudes of recombinant pox boosted CTL responses decayed rapidly and were comparable to those of the DNA-alone-vaccinated monkeys by the time of viral challenge. Consistent with these comparable memory T cell responses, the clinical protection seen in all groups of experimentally vaccinated monkeys was similar. This study, therefore, indicates that the steady-state memory, rather than the peak effector vaccine-elicited T lymphocyte responses, may be the critical immune correlate of protection for a CTL-based HIV vaccine.