This target article describes the current state of global gene replacement in the brain using viral vectors and assesses possible solutions to some of the many problems inherent in gene therapy of the central nervous system (CNS). Gene replacement therapy in the CNS is a potential means of producing a stable expression of normal human proteins in deficient cells and thus curing certain genetically inherited enzyme deficiencies and metabolic diseases as well as cancers. The two major issues to be addressed in CNS gene replacement are the delivery of genetic material to the brain and the expression of recombinant genetic material in target cells within the CNS. Focal inoculation of recombinant virions or other genetic vectors has limitations in global CNS disease. A new approach is the blood-brain barrier (BBB) disruption technique developed in this laboratory, in which hypertonic mannitol transiently shrinks the BBB endothelium, allowing passage of high molecular weight compounds and even viruses. Gene therapy of the CNS will require a viral vector system that allows long-term, nontoxic gene expression in neurons or glial cells. Retroviral vectors have limitations in CNS gene replacement, although they are suitable for expressing recombinant genes in intracerebral grafts, or toxic genes in brain tumors. Using mutant neurotropic viruses with reduced neurotoxicity (such as defective herpes simplex virus type I [HSV-1], the HSV-1 amplicon vector system we have developed, or adenovirus mutants) has potential for direct treatment of neurons. Injecting these vectors into rodent brains can lead to stable expression of foreign genetic material in postmitotic neuronal cells. We discuss our BBB disruption delivery technique, our defective HSV-1 amplicon vector system, and our feline model for the neuronal lysosomal storage disorder Gm2-gangliosidosis (Sandhoff disease), which may prove to be a useful model system for CNS gene therapy.