A significant obstacle to HIV vaccine development lies in the remarkable diversity of envelope proteins, the major targets of neutralizing antibody. That envelope diversity must be targeted is demonstrated by results from nonhuman primate studies in which single-envelope vaccines have protected against homologous, but rarely against heterologous virus challenges. Similarly, in clinical trials, single-envelope vaccines have failed to prevent break-through infections when challenge viruses were inevitably mismatched with the vaccine. To protect humans from infection by any isolate of HIV, we have prepared vaccine cocktails combining multiple envelopes from distinct viral isolates. We have tested several vehicles for vaccine delivery in small animals and have shown that successive immunizations with envelope, presented first as a DNA recombinant, then as a vaccinia virus (VV) recombinant, and finally as purified protein elicited strong neutralizing antibody responses. We have also tested the VV recombinant vaccine in chimpanzees. Pairs of animals received either single- or multi-envelope VV recombinant vaccines administered by the subcutaneous route. Results showed that the multi-envelope vaccine was safe, immunogenic, and superior to the single-envelope vaccine in eliciting HIV-specific antibody measurable in a standard clinical, immune assay. The promise of this system has led to the initiation of clinical trials, with which the hypothesis that cocktail vaccines will prevent human HIV infections may ultimately be tested.