Perhaps as many as 25-50% of adult patients and children with acquired immunodeficiency syndrome (AIDS) eventually suffer from neurological manifestations, including dysfunction of cognition, movement, and sensation. How can human immunodeficiency virus type 1 (HIV-1) result in neuronal damage if neurons themselves are for all intents and purposes not infected by the virus? This article reviews a series of experiments leading to a hypothesis that accounts at least in part for the neurotoxicity observed in the brains of AIDS patients. There is growing support for the existence of HIV- or immune-related toxins that lead indirectly to the injury or demise of neurons via a potentially complex web of interactions among macrophages (or microglia), astrocytes, and neurons. HIV-infected monocytoid cells (macrophages, microglia, or monocytes), after interacting with astrocytes, secrete eicosanoids, i.e., arachidonic acid and its metabolites, including platelet-activating factor. Macrophages activated by HIV-1 envelope protein gp120 also appear to release arachidonic acid and its metabolites. In addition, interferon-gamma (IFN-gamma) stimulation of macrophages induces release of the glutamate-like agonist, quinolinate. Furthermore, HIV-infected macrophage production of cytokines, including TNF-alpha and IL1-beta, contributes to astrogliosis. A final common pathway for neuronal susceptibility appears to be operative, similar to that observed in stroke, trauma, epilepsy, neuropathic pain, and several neurodegenerative diseases, possibly including Huntington's disease, Parkinson's disease, and amyotrophic lateral sclerosis. This mechanism involves the activation of voltage-dependent Ca2+ channels and N-methyl-D-aspartate (NMDA) receptor-operated channels, and, therefore, offers hope for future pharmacological intervention. This article focuses on clinically tolerated calcium channel antagonists and NMDA antagonists with the potential for trials in humans with AIDS dementia in the near future.