Late infantile neuronal ceroid lipofuscinosis (LINCL), a fatal, lysosomal storage disorder caused by mutations in the CLN2 gene, results in a deficiency of tripeptidyl-peptidase I (TPP-I) activity in neurons. Our prior studies showed that delivery of the human CLN2 cDNA directly to the CNS, using an adeno-associated virus serotype 2 (AAV2) vector, is safe in children with LINCL. As a second-generation strategy, we have demonstrated that AAVrh.10hCLN2, a rhesus-derived AAV vector, mediates wide distribution of TPP-I through the CNS in a murine model. This study tests the hypothesis that direct administration of AAVrh.10hCLN2 to the CNS of rats and nonhuman primates at doses scalable to humans has an acceptable safety profile and mediates significant CLN2 expression in the CNS. A dose of 10^11 genome copies (GC) was administered bilaterally to the striatum of Sprague Dawley rats with sacrifice at 7 and 90 days with no significant impact except for mild vector-related histopathological changes at the site of vector administration. A dose of 1.8 x 10^12 GC of AAVrh.10hCLN2 was administered to the CNS of 8 African green monkeys. The vector-treated monkeys did not differ from controls in any safety parameter except for mild to moderate white matter edema and inflammation localized to the administration sites of the vector. There were no clinical sequelae to these localized findings. TPP-I activity was > 2 SD over background in 31.7 +/- 8.1% of brain at 90 days. These findings establish the dose and safety profile for human clinical studies for the treatment of LINCL with AAVrh.10hCLN2.