Dominant optic atrophy (DOA) is an inherited mitochondrial disease leading to specific degeneration of retinal ganglion cells (RGCs), thus compromising transmission of visual information from the retina to the brain. Usually, DOA starts during childhood and evolves to poor vision or legal blindness, affecting the central vision, whilst sparing the peripheral visual field. In 20% of cases, DOA presents as syndromic disorder, with secondary symptoms affecting neuronal and muscular functions. Twenty years ago, we demonstrated that heterozygous mutations in OPA1 are the most frequent molecular cause of DOA. Since then, variants in additional genes, whose functions in many instances converge with those of OPA1, have been identified by next generation sequencing. OPA1 encodes a dynamin-related GTPase imported into mitochondria and located to the inner membrane and intermembrane space. The many OPA1 isoforms, resulting from alternative splicing of three exons, form complex homopolymers that structure mitochondrial cristae, and contribute to fusion of the outer membrane, thus shaping the whole mitochondrial network. Moreover, OPA1 is required for oxidative phosphorylation, maintenance of mitochondrial genome, calcium homeostasis and regulation of apoptosis, thus making OPA1 the Swiss army-knife of mitochondria. Understanding DOA pathophysiology requires the understanding of RGC peculiarities with respect to OPA1 functions. Besides the tremendous energy requirements of RGCs to relay visual information from the eye to the brain, these neurons present unique features related to their differential environments in the retina, and to the anatomical transition occurring at the lamina cribrosa, which parallel major adaptations of mitochondrial physiology and shape, in the pre- and post-laminar segments of the optic nerve. Three DOA mouse models, with different Opa1 mutations, have been generated to study intrinsic mechanisms responsible for RGC degeneration, and these have further revealed secondary symptoms related to mitochondrial dysfunctions, mirroring the more severe syndromic phenotypes seen in a subgroup of patients. Metabolomics analyses of cells, mouse organs and patient plasma mutated for OPA1 revealed new unexpected pathophysiological mechanisms related to mitochondrial dysfunction, and biomarkers correlated quantitatively to the severity of the disease. Here, we review and synthesize these data, and propose different approaches for embracing possible therapies to fulfil the unmet clinical needs of this disease, and provide hope to affected DOA patients.