Adrenomyeloneuropathy (AMN) is a late-onset axonopathy of spinal cord tracts caused by mutations of the ABCD1 gene that encodes adrenoleukodystrophy protein (ALDP), a peroxisomal transporter of very long-chain fatty acids (VLCFA). Disturbed metabolic interaction between oligodendrocytes (OL) and axons is suspected to play a major role in AMN axonopathy. To develop a vector targeting OL, the human ABCD1 gene driven by a short 0.3 kb part of the human myelin-associated glycoprotein (MAG) promoter was packaged into an adeno-associated viral serotype 9 (rAAV9). An intravenous injection of this vector on postnatal day 10 in Abcd1-/- mice, a model of AMN, allowed a near normal motor performance to persist for 24 months, while age-matched untreated mice developed major defects of balance and motricity. Three weeks postvector, 50-54% of spinal cord white matter OL was expressing human ALDP (hALDP) at the cervical level, and only 6-7% after 24 months. In addition, 29-32% of cervical spinal cord astrocytes at 3 weeks and 16-19% at 24 months also expressed ALDP. C26:0-lysoPC, a sensitive VLCFA marker of AMN, was lower by 41% and 50%, respectively, in the spinal cord and brain of vector-treated compared with untreated mice. In a nonhuman primate, the intrathecal injection of the rAAV9-MAG vector induced abundant ALDP expression at 3 weeks in spinal cord OL (43%, 29%, and 26% at cervical, thoracic, and lumbar levels) and cerebellum OL (35%). In addition, 33-41% of spinal cord astrocytes expressed hALDP, and 27% of cerebellar astrocytes. To our knowledge, OL targeting had not been obtained before in primates with other vectors or promoters. The current results thus provide a robust proof-of-concept not only for the gene therapy of AMN but also for other central nervous system diseases, where the targeting of OL with the rAAV9-MAG vector may be of interest.