Adeno-associated virus (AAV) is a promising vector for central nervous system (CNS) gene transfer, but a number of issues must be addressed if AAV is to be used for widespread delivery throughout the CNS. Our aim was to test the effect of dose, route of delivery, and hydroxyurea treatment on brain expression of beta-galactosidase activity after cerebral inoculation with an rAAV-lacZ vector (rAAV-beta-gal). We also wished to test whether an immune response appeared against the vector and the transgene product. We found in BALB/c mice that beta-Gal expression increased during the first 2 months after inoculation, then decreased slightly by 4 months, and continued out to 6, 12, and 15 months in single animals. Cerebral injection produced localized beta-Gal expression that did not diffuse to other regions despite a fivefold increase in injection volume. Intraventricular injection resulted in negligible transduction. Antibodies to AAV capsid protein and beta-Gal appeared at low levels at 2 and 4 months, but correlated poorly with beta-Gal expression and did not prevent readministration of rAAV-beta-gal. Hydroxyurea treatment did not result in increased transduction in vivo. We conclude that our study confirms rAAV vectors as having considerable potential for CNS gene transfer; however, several important problems must be addressed if this vector system is to be used for long-term transduction of the entire brain. Sustained, regulatable expression will be needed if rAAV is to be used in the treatment of chronic CNS disease. The difficulty in delivering AAV to diverse regions of the brain is an important problem that must be overcome if these vectors are to be used for anything beyond localized transduction.