HIV invades the brain during acute infection. Yet, it is unknown whether long-lived infected brain cells release productive virus that can egress from the brain to re-seed peripheral organs. This understanding has significant implication for the brain as a reservoir for HIV and most importantly HIV interplay between the brain and peripheral organs. Given the sheer number of astrocytes in the human brain and their controversial role in HIV infection, we evaluated their infection in vivo and whether HIV infected astrocytes can support HIV egress to peripheral organs. We developed two novel models of chimeric human astrocyte/human peripheral blood mononuclear cells: NOD/scid-IL-2Rgc null (NSG) mice (huAstro/HuPBMCs) whereby we transplanted HIV (non-pseudotyped or VSVg-pseudotyped) infected or uninfected primary human fetal astrocytes (NHAs) or an astrocytoma cell line (U138MG) into the brain of neonate or adult NSG mice and reconstituted the animals with human peripheral blood mononuclear cells (PBMCs). We also transplanted uninfected astrocytes into the brain of NSG mice and reconstituted with infected PBMCs to mimic a biological infection course. As expected, the xenotransplanted astrocytes did not escape/migrate out of the brain and the blood brain barrier (BBB) was intact in this model. We demonstrate that astrocytes support HIV infection in vivo and egress to peripheral organs, at least in part, through trafficking of infected CD4+ T cells out of the brain. Astrocyte-derived HIV egress persists, albeit at low levels, under combination antiretroviral therapy (cART). Egressed HIV evolved with a pattern and rate typical of acute peripheral infection. Lastly, analysis of human cortical or hippocampal brain regions of donors under cART revealed that astrocytes harbor between 0.4-5.2% integrated HIV gag DNA and 2-7% are HIV gag mRNA positive. These studies establish a paradigm shift in the dynamic interaction between the brain and peripheral organs which can inform eradication of HIV reservoirs.