Mitochondrial diseases (MDs) may result from mutations affecting nuclear or mitochondrial genes, encoding mitochondrial proteins, or non-protein-coding mitochondrial RNA. Despite the great variability of affected genes, in the most severe cases, a neuromuscular and neurodegenerative phenotype is observed, and no specific therapy exists for a complete recovery from the disease. The most used treatments are symptomatic and based on the administration of antioxidant cocktails combined with antiepileptic/antipsychotic drugs and supportive therapy for multiorgan involvement. Nevertheless, the real utility of antioxidant cocktail treatments for patients affected by MDs still needs to be scientifically demonstrated. Unfortunately, clinical trials for antioxidant therapies using α-tocopherol, ascorbate, glutathione, riboflavin, niacin, acetyl-carnitine and coenzyme Q have met a limited success. Indeed, it would be expected that the employed antioxidants can only be effective if they are able to target the specific mechanism, i.e., involving the central and peripheral nervous system, responsible for the clinical manifestations of the disease. Noteworthily, very often the phenotypes characterizing MD patients are associated with mutations in proteins whose function does not depend on specific cofactors. Conversely, the administration of the antioxidant cocktails might determine the suppression of endogenous oxidants resulting in deleterious effects on cell viability and/or toxicity for patients. In order to avoid toxicity effects and before administering the antioxidant therapy, it might be useful to ascertain the blood serum levels of antioxidants and cofactors to be administered in MD patients. It would be also worthwhile to check the localization of mutations affecting proteins whose function should depend (less or more directly) on the cofactors to be administered, for estimating the real need and predicting the success of the proposed cofactor/antioxidant-based therapy.