Substantial effort is being directed toward generating vaccines that can prevent the heterosexual transmission of HIV-1. If "selection" for specific variants during sexual intercourse occurs, then vaccines should be designed to prevent transmission of these specific viruses. Using the SIV-rhesus macaque model to test the hypothesis that specific HIV genotypes are more efficient at producing infection by sexual transmission, it was possible to demonstrate that the genotypic determinants that permit SIV or SHIV to produce systemic infection differ depending on the route of virus inoculation. This finding supports the conclusion that there is selection for viral genotypes during sexual transmission of HIV. However, the ability of a virus to grow in rhesus macaque monocyte-derived macrophages in vitro does not predict the outcome of intravaginal inoculation with that virus. We did find that after intravenous inoculation all the vaginally transmitting viruses produced plasma antigenemia and high levels of plasma viral RNA. In contrast, although the nontransmitting viruses infect rhesus macaques after intravenous inoculation, the infection that occurs after intravenous inoculation is characterized by a lack of viral antigen in plasma and low levels of plasma viral RNA. On the basis of these results, it is clear that viruses which are adapted to replicate to high levels in vivo are transmitted by vaginal inoculation. This principle may also apply to the transmission of HIV in humans.