The ability of antiviral and antiretroviral drugs to enter the brain is a critical issue in the treatment of many viral brain diseases, including HIV-related neurologic disease. Much of the literature concerning nucleoside analog entry into the nervous system focuses on drug levels in the cerebrospinal fluid (CSF), equating these with drug levels in the brain extracellular fluid (ECF) as though the two compartments intermix freely. We review the anatomic and physiologic aspects of drug entry into CSF and into brain ECF, as well as the exchange processes between these two compartments. In most instances drug concentrations in the CSF and ECF compartments bear little relationship to one another and using CSF concentrations to extrapolate brain ECF concentrations may significantly overestimate the latter. Accepted terminology and methodology for making measurements of blood-brain barrier function are discussed. Studies of brain uptake that express results as brain:plasma ratios, or that have used microdialysis, may overestimate the amount of drug reaching the brain. Using published data, we present an estimate of the time course of Zidovudine (AZT) concentrations in brain ECF and show that brain concentrations of AZT will likely be below that necessary to inhibit HIV-1 replication when AZT is administered systemically. Antiviral nucleosides and oligonucleotides appear to have limited entry into the brain when given systemically, which may hinder therapy of viral brain diseases, while some of the protease inhibitors may enter the brain more readily. Alternative methods for increasing antiviral and antiretroviral drug delivery to brain are discussed.