GV-TH-01, a Phase 1 open-label trial of a DNA prime - Modified Vaccinia Ankara (MVA) boost vaccine (GOVX-B11), was undertaken in HIV infected participants on antiretroviral treatment (ART) to evaluate safety and vaccine-elicited T cell responses, and explore the ability of elicited CD8+ T cells to control viral rebound during analytical treatment interruption (TI). Nine men who began antiretroviral therapy (ART) within 18 months of seroconversion and had sustained plasma HIV-1 RNA <50 copies/mL for at least 6 months were enrolled. Median age was 38 years, median pre-ART HIV-1 RNA was 140,000 copies/ml and mean baseline CD4 count was 755/μl. Two DNA, followed by 2 MVA, inoculations were given 8 weeks apart. Eight subjects completed all vaccinations and TI. Clinical and laboratory adverse events were generally mild, with no serious or grade 4 events. Only reactogenicity events were considered related to study drug. No treatment emergent viral resistance was seen. The vaccinations did not reduce viral reservoirs and virus re-emerged in all participants during TI, with a median time to re-emergence of 4 weeks. Eight of 9 participants had CD8+ T cells that could be stimulated by vaccine-matched Gag peptides prior to vaccination. Vaccinations boosted these responses as well as eliciting previously undetected CD8+ responses. Elicited T cells did not display signs of exhaustion. During TI, temporal patterns of viral re-emergence and Gag-specific CD8+ T cell expansion suggested that vaccine-specific CD8+ T cells had been stimulated by re-emergent virus in only 2 of 8 participants. In these 2, transient decreases in viremia were associated with Gag selection in known CD8+ T cell epitopes. We hypothesize that escape mutations, already archived in the viral reservoir, plus a poor ability of CD8+ T cells to traffic to and control virus at sites of re-emergence, limited the therapeutic efficacy of the DNA/MVA vaccine.