Anti-retroviral therapy (ART) has been highly successful in controlling HIV replication, reducing viral burden, and preventing both progression to AIDS and viral transmission. Yet, ART alone cannot cure the infection. Even after years of successful therapy, ART withdrawal leads inevitably to viral rebound within a few weeks or months. Our hypothesis: effective therapy must control both the replicating virus pool and the reactivatable latent viral reservoir. To do this, we have combined ART and immunotherapy to attack both viral pools simultaneously. The vaccine regimen consisted of DNA vaccine expressing SIV Gag, followed by a boost with live attenuated rubella/gag vectors. The vectors grow well in rhesus macaques, and they are potent immunogens when used in a prime and boost strategy. We infected rhesus macaques by high dose mucosal challenge with virulent SIVmac251 and waited three days to allow viral dissemination and establishment of a reactivatable viral reservoir before starting ART. While on ART, the control group received control DNA and empty rubella vaccine, while the immunotherapy group received DNA/gag prime, followed by boosts with rubella vectors expressing SIV gag over 27 weeks. Both groups had a vaccine "take" to rubella, and the vaccine group developed antibodies and T cells specific for Gag. Five weeks after the last immunization, we stopped ART and monitored virus rebound. All four control animals eventually had a viral rebound, and two were euthanized for AIDS. One control macaque did not rebound until 2 years after ART release. In contrast, there was only one viral rebound in the vaccine group. Three out of four vaccinees had no viral rebound, even after CD8 depletion, and they remain in drug-free viral remission more than 2.5 years later. The strategy of early ART combined with immunotherapy can produce a sustained SIV remission in macaques and may be relevant for immunotherapy of HIV in humans.