Mutations affecting mitochondrial function lead to a variety of disorders affecting the heart, skeletal muscles, endocrine glands and central nervous system. These diseases have been hard to replicate in animals, because of the difficulty in manipulating mitochondrial DNA and the lethality of nuclear mutations affecting mitochondrial respiration. We have used an AAV-mediated RNA knockdown strategy to produce two mouse models of Leber Hereditary Optic Neuropathy (LHON). Patients with LHON typically contain mutations in one of the subunits of Complex I of the mitochondrial respiratory chain (NADH-ubiquinone oxidoreductase) and experience bilateral loss of vision beginning around age 20. In mice, blocking synthesis of subunits of this enzyme leads to the pathologic hallmarks of the disease—apoptotic death of retinal ganglion cells, demyelination and loss of axons in the optic nerve. Using these mouse models, we have tested two approaches for gene therapy of LHON. In the first strategy, we delivered the gene for mitochondrial manganese superoxide dismutase (SOD2) to limit the accumulation of reactive oxygen species that result from Complex I defects. Delivery of SOD2 caused an 18% increase (p=0.001) in the survival of retinal ganglion cells and a proportional increase in the thickness of the optic nerve. In the second therapeutic strategy, we re-coded the mitochondrial gene for the ND4 subunit of Complex I, attached a mitochondrial transport sequence to its amino terminus and expressed it as a nuclear gene (allotopic expression). Treatment of the optic nerve with AAV expressing this protein led to a 13% increase (p<0.01) in diameter of the optic nerve relative to contralateral eyes treated with a control virus. Thus both approaches prevented tissue injury resulting from defects to Complex I of the respiratory chain. These provide the first evidence in animals that gene therapy for mitochondrial disease may be possible.