Viral vectors have recently been used successfully to transfer genes and express different proteins in the brain. This review discusses the requirements to consider human clinical trials in which recombinant adeno-associated virus vectors are used to transfer the genes necessary to produce l-dihydroxyphenylalanine (l-dopa) directly into the striatum of Parkinson's patients. Preclinical data that apply to the criteria defined as prerequisite for clinical trials are discussed. Thus, in animal models using recombinant adeno-associated virus vectors it has been demonstrated that l-dopa can be synthesized in the striatum after in vivo transduction. In addition, these l-dopa levels are sufficient to affect behavior in a dopamine-deficient animal model, the expression is extremely long-lasting, and the ability to transcriptionally regulate tyrosine hydroxylase has been demonstrated but not fully characterized. However, while immune responses to recombinant adeno-associated virus infection in the periphery have been studied, direct assessment of the potential immune response in the brain has not been sufficiently defined. Therefore, the rationale for delivering l-dopa directly to the striatum to treat Parkinson's disease is sound and the preclinical data are promising but all the issues surrounding this strategy are not resolved.