Mucopolysaccharidosis type I (MPS I) is a lysosomal storage disease resulting from deficiency of the enzyme α-L-iduronidase (IDUA). Accumulation of the IDUA substrates heparan and dermatan sulfate causes widespread organ pathology. While many of the somatic manifestations of MPS I can be treated with intravenous enzyme replacement, the devastating CNS sequelae—cognitive impairment, spinal cord compression, and hydrocephalus—do not respond to treatment. Partial preservation of cognitive function is possible with early hematopoietic stem cell transplantation, although transplant is associated with substantial morbidity and mortality. Gene transfer using adeno-associated virus (AAV) vectors offers a potential alternative approach to deliver the IDUA enzyme to the CNS. Introducing a functional IDUA gene to a subset of quiescent cells could provide a permanent source of secreted enzyme beyond the blood-brain barrier. However, preclinical studies evaluating direct injection of AAV vectors into the brain have shown that transduction is limited to small regions surrounding the injection site, and that injection can be associated with a localized inflammatory response and immune-mediated killing of transduced cells. In order to overcome these limitations, we evaluated delivery of an AAV serotype 9 vector into the cerebrospinal fluid as a less invasive method to achieve widespread brain transduction. Studies in nonhuman primates demonstrated that intrathecal AAV9 delivery results in transduction of cells throughout the brain and spinal cord without eliciting destructive immune responses to the transgene product. Intrathecal injection of AAV9 vectors expressing IDUA in canine and feline models of MPS I replicated the widespread transduction observed in primate studies, and demonstrated resolution of storage lesions throughout the CNS. Antibodies against the enzyme were detected in the CSF of some animals, which coincided with lower CSF IDUA activity and less efficient correction of storage lesions. We found that immunological tolerance could be induced to IDUA by exposing newborn MPS I dogs to the enzyme, which enhanced the efficacy of subsequent gene transfer. These results were replicated in rhesus macaques, supporting the potential to translate neonatal tolerance induction to clinical applications. Intrathecal AAV delivery offers the potential for widespread gene transfer in the CNS with a single minimally invasive vector injection, which could prove transformative for the field of gene therapy for inherited neurological disorders.