The ultimate goal in the treatment of HIV-infected persons is to prevent disease progression. A strategy to accomplish this goal is to use chemotherapy to reduce viral load followed by immunotherapy to stimulate HIV-specific immune responses that are observed in long-term asymptomatic individuals. An effective, live, recombinant virus, expressing HIV sequences, would be capable of inducing both CTL and CD4(+) helper T cell responses. To accomplish these goals, the viral vector must be immunogenic yet retain its avirulent phenotype in a T cell-deficient host. We have identified a coxsackievirus variant, CB4-P, that can induce protective immunity against a virulent variant. In addition, the CB4-P variant remains avirulent in mice lacking CD4(+) helper T cells, suggesting that CB4-P may be uniquely suited as a viral vector for a therapeutic HIV vaccine. Two strategies designed to elicit CTL and CD4(+) helper T cell responses were used to construct CB4-P/HIV recombinants. Recombinant viruses were viable, genetically stable, and retained the avirulent phenotype of the parental virus. In designing a viral vector for vaccine development, an issue that must be addressed is whether preexisting immunity to the vector would affect subsequent administration of the recombinant virus. Using a test recombinant, we showed that prior exposure to the parental CB4-P virus did not affect the ability of the recombinant to induce a CD4(+) T cell response against the foreign sequence. The results suggest that a "cocktail" of coxsackie/HIV recombinants may be useful as a therapeutic HIV vaccine.