A variety of HIV-induced lesions of the central nervous system (CNS) have been described, including HIV encephalitis, HIV leukoencephalopathy, axonal damage, and diffuse poliodystrophy with neuronal loss of variable severity resulting, at least partly, from an apoptotic process. However, no correlation could be established between these changes and HIV dementia (HIVD). From our study of HIV infected patients, it appeared that neuronal apoptosis is probably not related to a single cause. Microglial and glial activation, directly or indirectly related to HIV infection, plays a major role in neuronal apoptosis possibly through the mediation of oxidative stress. In our patients with full-blown AIDS, this mechanism predominated in the basal ganglia and correlated well with HIVD. Axonal damage, either secondary to microglial activation, or to systemic factors also contributes to neuronal apoptosis. Although massive neuronal loss may be responsible for HIVD in occasional cases, we conclude that neuronal apoptosis is a late event and does not represent the main pathological substrate of HIVD. The dementia more likely reflects a specific neuronal dysfunction resulting from the combined effects of several mechanisms, some of which may be reversible. Introduction of highly active antiretroviral therapy dramatically improved patient survival, however, its impact on the incidence and course of HIVD remains debatable. In our series, the incidence of HIVE has dramatically decreased since the introduction of multitherapies, but a number of cases remain whose cognitive disorders persist, despite HAART. The poor CNS penetration of many antiretroviral agents is a possible explanation, but irreversible "burnt out" HIV-induced CNS changes may also be responsible.