Nonhuman primate-simian immunodeficiency virus (SIV) models are powerful tools for studying the pathogenesis of human immunodeficiency virus type 1 (HIV-1) in the brain. Our laboratory recently isolated a neuropathogenic viral swarm, SIVsmH804E, a derivative of SIVsmE543-3, which was the result of sequential intravenous passages of viruses isolated from the brains of rhesus macaques with SIV encephalitis. Animals infected with SIVsmH804E or its precursor (SIVsmH783Br) developed SIV meningitis and/or encephalitis at high frequencies. Since we observed macaques with a combination of meningitis and encephalitis, as well as animals in which meningitis or encephalitis was the dominant component, we hypothesized that distinct mechanisms could be driving the two pathological states. Therefore, we assessed viral populations in the meninges and the brain parenchyma by laser capture microdissection. Viral RNAs were isolated from representative areas of the meninges, brain parenchyma, terminal plasma, and cerebrospinal fluid (CSF) and from the inoculum, and the SIV envelope fragment was amplified by PCR. Phylogenetic analysis of envelope sequences from the conventional progressors revealed compartmentalization of viral populations between the meninges and the parenchyma. In one of these animals, viral populations in meninges were closely related to those from CSF and shared signature truncations in the cytoplasmic domain of gp41, consistent with a common origin. Apart from magnetic resonance imaging (MRI) and positron-emission tomography (PET) imaging, CSF is the most accessible assess to the central nervous system for HIV-1-infected patients. However, our results suggest that the virus in the CSF may not always be representative of viral populations in the brain and that caution should be applied in extrapolating between the properties of viruses in these two compartments.