Many neurological disorders result directly or indirectly from the loss of inhibitory function. Engineering the production of GABA, an inhibitory neurotransmitter, may therefore be able at least partly to restore the lost inhibition seen in epilepsy, Parkinson's disease, or Huntington's disease. In this article, we describe a set of recombinant adeno-associated viruses (AAVs) that can deliver cDNAs encoding the GABA-producing enzyme, glutamate decarboxylase (GAD), directly into neural cells. We have characterized these recombinant AAVs in several cell lines derived from the CNS. These recombinant AAVs effectively transduced all neural cell lines, although with different efficiencies. Transduction occurred in both proliferating and nonproliferating cells, but actively proliferating cell lines had approximately six times greater transduction efficiency than nonproliferating cells. Furthermore, these AAVs maintained long-term expression of GAD in an astrocytic cell line for at least seven passages. These recombinant AAVs are promising vehicles for investigating the potential therapeutic effects of GABA in animal models of epilepsy and neurodegenerative diseases.