Optic neuropathies such as Leber's hereditary optic neuropathy, dominant optic atrophy and toxic amblyopia are an important cause of irreversible visual failure. Although they are associated with a defect of mitochondrial energy production, their pathogenesis is poorly understood. A common feature to all these disorders is relatively selective degeneration of the papillomacular bundle of retinal ganglion cells resulting central or caecocentral visual field defects. The striking similarity in the pattern of clinical involvement seen with these disparate disorders suggests a common pathway in their aetiology. The existing hypothesis that the optic nerve head has higher energy demands than other tissues making it uniquely dependent on oxidative phosporylation is not satisfactory. First, other ocular tissues such as photoreceptors, which are more dependent on oxidative phosporylation are not affected. Second, other mitochondrial disorders, which have a greater impact on mitochondrial energy function, do not affect the optic nerve. The optic nerve head has certain unique ultra structural features. Ganglion cell axons exit the eye through a perforated collagen plate, the lamina cribrosa. There is a sharp discontinuity in the density of mitochondria at the optic nerve head, with a very high concentration in the prelaminar nerve fibre layer and low concentration behind the lamina. This has previously been attributed to a mechanical hold up of axoplasmic flow, which has itself been proposed as a factor in the pathogenesis of a number of optic neuropathies. More recent evidence shows that mitochondrial distribution reflects the different energy requirements of the unmyelinated prelaminar axons in comparison to the myelinated retrolaminar axons. The heterogeous distribution of mitochondria is actively maintained to support conduction through the optic nerve head. We propose that factors that disrupt the heterogeneous distribution of mitochondria can result in ganglion cell death. Evidence for this comes from studies of cultured cells with the dominant optic atrophy mutation in which mitochondrial distribution is altered and from some forms of hereditary spastic paraparesis which are associated with optic atrophy. The responsible mutations do not affect ATP production until late in the disease but do affect mitochondrial arrangement, again showing that mitochondrial distribution as well as energy production by individual mitochondria may be important in the pathogenesis of ganglion cell death. Greater understanding of the factors localising mitochondria within the ganglion cell axon in particular the interaction with cytoskeleton is required to formulate new treatments. Boosting energy production alone may not be an effective treatment.