Replication of HIV-1 within the CNS compartment can only be bad. At a minimum the detection by the immune system of replicating virus within the CNS leads to a neuro-toxic inflammatory response. If the clearance of virus is incomplete then a persistent inflammatory response could provide the basis for sustained cell damage within the CNS. Markers of viral replication in the CNS in living subjects must come from the analysis of virus in the CSF, or be inferred from indirect markers from non-invasive imaging techniques or from cell or inflammatory markers in the CSF or blood. Virus can be found in the CSF as the result of three distinct mechanisms. Infected T cells traffic into the CNS and release virus that by sequence looks just like the virus in the blood, and this virus can be present at a low level or, with pleocytosis, at a high level. Alternatively, some of this virus can be transiently clonally amplified by a poorly understood mechanism to give elevated viral loads but of surprisingly homogeneous viral populations. Finally, virus can take root within the CNS for sustained replication but in an environment that is poor in CD4+ T cells. This causes the virus to adapt to macrophage-like cells with their low levels of the surface CD4 protein needed for entry. The unique genetic lineage of virus replicating independently in the CNS and the phenotypic adaptation to this CD4 low environment provide clear evidence of a distinct viral population and site of ongoing viral replication. Therapy suppresses viral load in the blood and the CSF. However, antiviral drug levels are reduced in the CNS, a situation that would result in viral replication and the selection of resistance if it took place in the blood. In about 5-10% of subjects on therapy there is detectable virus in the CSF either in the absence of or at levels higher than virus in the blood. This CSF escape virus takes on several forms, including a complex viral population at least partially adapted to a CD4 low environment. We do not know how well the CSF records virologic events that may be ongoing deep within the brain parenchyma. Such replication when detected in the CSF represents evidence of an active viral reservoir on therapy, while the potential for a unique latent reservoir within the CNS is of considerable interest but challenging to study.