Mitochondrial disease is a group of rare conditions, with no approved treatment to date, except for Leber hereditary optic neuropathy. Therapeutic options to alleviate the symptoms of mitochondrial disease are urgently needed. Sonlicromanol is a promising candidate, as it positively alters the key metabolic and inflammatory pathways associated with mitochondrial disease. Sonlicromanol is a reductive and oxidative distress modulator, selectively inhibiting microsomal prostaglandin E1 synthase activity. This Phase 2b program, aiming at evaluating sonlicromanol in adults with m.3243A>G mutation and primary mitochondrial disease, consisted of a randomized controlled (RCT) study (dose-selection) followed by a 52-week open-label extension study (EXT, long-term tolerability, safety, and efficacy of sonlicromanol). Patients were randomized (1:1:1) to receive 100- or 50-mg sonlicromanol, or placebo twice daily (bid) for 28 days with ≥2-week wash-out period between treatments. Patients who completed the RCT study entered the EXT study wherein they received 100-mg sonlicromanol bid. Overall, 27 patients were randomized (24 RCT patients completed all periods). 15 patients entered the EXT, and 12 patients were included in the EXT analysis set. All patients reported good tolerability and favourable safety, with pharmacokinetic results comparable to the earlier Phase 2a study. The RCT primary endpoint (change from placebo in the attentional domain of cognition score [IDN: visual identification, Cogstate]) did not reach statistical significance. Using a categorisation of the subject's period baseline a treatment effect over placebo was observed if their baseline was more affected (p=0.0338). Using this approach, there were signals of improvements over placebo in at least one dose in the Beck Depression Inventory (BDI, p=0.0143), Cognitive Failure Questionnaire (CFQ, p=0.0113), and the Depression subscale of the Hospital Anxiety and Depression Scale (p=0.0256). Statistically and/or clinically meaningful improvements were observed in the patient- and clinician-reported outcome measures at the end of the EXT study (Test of attentional performance [TAP] with alarm, p=0.0102; TAP without alarm, p=0.0047; BDI somatic, p=0.0261; BDI Total, p=0.0563; SF12 physical component score, p=0.0008). Seven of nine domains of RAND-Short form-36 like SF-36 pain improved (p=0.0105). Other promising results were observed in Neuro QoL-Fatigue-SF (p=0.0036), MiniBESTest (p=0.0009), McGill Pain Questionnaire (p=0.0105), EQ-5D-5L-VAS (p=0.0213) and EQ-5D-5L-index (p=0.0173). Most patients showed improvement in the 5× sit-to-stand test. Sonlicromanol was well-tolerated and demonstrated a favourable benefit/risk ratio for up to one year. Sonlicromanol was efficacious in patients when affected at baseline, as seen across a variety of clinically relevant domains. Long-term treatment showed more pronounced changes from baseline.