There has been considerable debate over what evidence from preclinical and clinical studies is required to advance an HIV vaccine candidate to phase III efficacy testing. Given this situation, conduct of intermediate-size trials is proposed as a method for assessing the plausibility that a vaccine candidate would prevent chronic HIV infection. Designed to observe 45 incident infections in the control group, these preliminary efficacy trials could rule out candidates with low or no efficacy while advancing those candidates with some evidence of protection to definitive trials. In addition, these trials could provide clues about correlates of immunity. A threefold or greater difference in the postvaccination geometric mean titer of neutralizing antibody can be readily detected between infected and uninfected vaccinees. Differences in CD8+ cytotoxic T lymphocytes, however, are more difficult to detect. Intermediate-size trials could also discern a 0.5 log10 or greater difference in plasma HIV-1 RNA levels between infected vaccinees and infected controls. Such differences in viral load might suggest disease amelioration or reduction in infectiousness. Given the large variability in CD4 count and its relatively modest average decline in the year after infection, a slower decline in CD4 count among infected vaccinees would not be detectable. With limited resources, intermediate-size trials could contribute significantly to HIV vaccine development.