Mitochondrial dysfunction
Mitochondria are the cell's power plants, and as they age they make less energy and spill their own DNA in a way the body mistakes for an infection — one of the leading suspected drivers of age-related inflammation. The twist is that they respond to stress: exercise does more for them than any supplement, because a controlled dose of stress is exactly what tells them to renew — and fasting, heat and cold are thought to work the same way.
Mitochondrial dysfunction is the seventh of the twelve hallmarks of aging, and one of the most connected — it both feeds and is fed by genomic instability, senescence, and inflammation. Mitochondria are tiny compartments inside nearly every cell that burn fuel and oxygen to make adenosine triphosphate (ATP), the molecule that powers almost everything a cell does. They are far more than batteries, though: they buffer calcium, decide when a damaged cell should die, and manage the production of reactive oxygen species (ROS) — the unstable, oxygen-containing by-products of energy generation that damage whatever they touch.[1] With age, the systems that keep this machinery in good repair break down, dysfunctional mitochondria pile up, and the consequences ripple out across the whole body.[2]
How confident should you be
Descriptive — Strong. Intervention — Moderate.
That mitochondria decay with age, and how, is settled. The link to inflammation and frailty in people is observational: in older adults, mitochondrial DNA circulating in the blood tracks inflammatory markers, but in a study of 672 older adults it did not predict death once inflammation was taken into account.[3] The intervention side is stronger here than for most hallmarks — exercise reliably and measurably improves mitochondrial function in humans, which is more than can be said for the levers aimed at genomic instability or proteostasis — but it stops short of proof that improving mitochondria extends life. There is no human lifespan data for any of it, the muscle–brain result that anchors the clinical case is observational, and no supplement in this space has shown a benefit on a hard outcome — the few trials that met their main goal were small, short and measured stand-ins such as insulin sensitivity or artery function, while the larger muscle and performance trials mostly missed theirs. The lifestyle levers are the real content of this page; the pills are a footnote to them.
Why mitochondria fail with age
Strong for the mitochondrial-DNA damage, which is mapped in aging human tissue; Moderate for the quality-control decline below, which rests largely on cell and animal work.
Mitochondria carry their own small loop of DNA, a relic of their ancient origin as free-living bacteria, and that DNA is unusually exposed. It is packed differently from the genome in the nucleus — in small protein bundles called nucleoids rather than wound around histones — it sits right next to the energy machinery, and its repair toolkit is more limited. It picks up mutations and deletions faster (one estimate puts the rate up to 15 times higher), and these have been mapped in aging human brain, heart, colon and muscle.[4] Why they build up is less simple than the classic free-radical story: in one ultra-sensitive sequencing study of human brain tissue, point mutations rose about five-fold over 80 years, yet the mutation type that oxidative damage leaves behind did not rise — suggesting most come from copying errors and spontaneous chemical decay of DNA rather than from free radicals.[5]
The cell's defence is a constant quality-control cycle that keeps the mitochondrial pool healthy. Mitochondria are rebuilt continuously, on a timescale of days to weeks depending on the tissue — in mice, the average mitochondrial protein has a half-life of about 4 days in the liver and about 17 days in the heart[6] — through three coordinated processes:
- Biogenesis — building new mitochondria, driven by a master switch called PGC-1α (the same switch endurance exercise flips on).
- Mitophagy — selectively digesting worn-out mitochondria; a sensor system tags any unit that loses its electrical charge for recycling.
- Fission and fusion — mitochondria constantly split apart and merge, which lets the network quarantine a damaged segment and dilute scattered mutations.
In youth this cycle keeps the network efficient. With age the balance tips: the biogenesis switch weakens, clearance slows, damaged mitochondria pile up, and the pool as a whole makes less ATP and leaks more ROS — a self-reinforcing decline, though most of the detail comes from cell and animal studies.[7]
From leaky mitochondria to whole-body inflammation
Strong for the mechanism; Moderate for its weight in human aging. The pathway is well characterised in cells and animals; how much of human inflammaging runs through it is not quantified.
The most important recent insight is how a local energy problem becomes a systemic one — and it runs through the same cytosolic-DNA alarm that connects genomic instability to aging. Because mitochondrial DNA still resembles bacterial DNA, the cell treats it as a threat if it ends up in the wrong place. When a damaged mitochondrion's membrane ruptures, its DNA spills into the cytoplasm, where the cGAS-STING sensor — the cell's burglar alarm for misplaced DNA — mistakes it for a virus and switches on a powerful inflammatory and antiviral program.[8]
A brief burst of this is useful housekeeping. The problem is chronic, low-grade activation that never resolves — an "inflammatory clock" that translates accumulated organelle damage into the persistent, sterile inflammation of inflammaging.[9] In senescent cells, sustained cGAS-STING signalling pours out the inflammatory secretions (the senescence-associated secretory phenotype) that damage neighbouring tissue.[10] This is the leading explanation for how failing mitochondria could drive body-wide aging rather than just a per-cell energy shortfall — well mapped in cells and mice, but not yet measured as a cause in people.
The muscle–brain axis: how mitochondria show up as frailty and dementia
Weak — observational. One prospective cohort for the brain link, plus cross-sectional studies of muscle samples. It establishes that muscle mitochondrial function tracks frailty and predicts cognitive decline, not that improving it prevents either.
Because some tissues are far more energy-hungry than others, mitochondrial decline tends to surface first where the demand is highest — skeletal muscle and the brain.
In muscle, failing mitochondria are a likely contributor to sarcopenia (age-related muscle loss) and frailty, though in people the evidence is association, not proof of cause. Muscle samples from 119 older men in Singapore, the UK and Jamaica showed fewer and weaker mitochondria in those with sarcopenia than in men of the same age without it.[11] In the largest study to measure this directly, about 900 people aged 70 and over had thigh-muscle samples tested: those with weaker muscle mitochondria were more likely to be frail and took longer to walk 400 metres, though the walking link ran largely through leg power and the energy cost of walking.[12][13] Strikingly, muscle mitochondrial health also tracks the brain. In the Baltimore Longitudinal Study of Aging, mitochondrial function in the thigh muscle — measured non-invasively by magnetic-resonance spectroscopy — predicted what happened in the brain. Participants whose muscle mitochondria scored one standard deviation better than the cohort average went on to develop mild cognitive impairment or dementia at roughly half the rate of the rest — a 52% lower hazard per standard deviation — and had about 59% lower odds, again per standard deviation, of a positive amyloid brain scan.[14] Read it as an association: people with better muscle mitochondria differ from those without in many ways, and no trial has tested whether improving the muscle measure changes the brain outcome. People with early mild cognitive impairment also show measurable differences in muscle mitochondrial respiratory capacity, underscoring that this is a systemic, not purely neurological, problem — though that comparison rests on only 50 people, and the effect was entangled with whether participants were taking the Alzheimer's drug donepezil.[15]
A related energy failure may underlie post-infectious chronic fatigue syndrome, though the evidence is early. One US National Institutes of Health study found raised levels of a stress-induced protein, WASF3, in muscle samples from 14 people with myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) compared with 10 healthy volunteers; in one patient's cells and in mice, the same protein jammed the assembly of the energy machinery and cut energy output.[16] The authors suggest long COVID might work the same way, but they studied no long-COVID patients, and the finding has not yet been independently confirmed. The broader association between mitochondrial dysfunction and chronic fatigue long predates COVID.[17]
Mitohormesis: a little stress is the point
Moderate. The principle is well established in cells and animals; the dose–response boundaries in humans are not mapped.
Here is the reframe that organises everything practical about this hallmark. The old "free-radical theory" cast all mitochondrial stress as damage to be minimised. The modern picture is mitohormesis: while high, chronic mitochondrial stress is destructive, a brief, controlled dose of stress triggers an adaptive overcompensation — the cell clears damaged mitochondria, builds new ones, and ends up more resilient than before.[18] Whether mitochondrial change is harmful or protective depends entirely on the dose.[19]
This has a counterintuitive practical consequence that is worth stating plainly, because it inverts a generation of supplement marketing. High-dose antioxidant supplements do not slow aging, and taken around training they can blunt part of what the training does. The transient burst of reactive oxygen species that exercise produces is not just a side effect to be mopped up — it is part of the signal that tells the cell to build more and better mitochondria.[20] In small trials of young adults, about 1 g of vitamin C a day together with vitamin E, taken during training, blunted the rise in muscle mitochondrial markers and, in one trial, cancelled the improvement in insulin sensitivity.[21][22] Measured fitness and strength have mostly come out the same: in the largest trial (54 people, 11 weeks), fitness rose by about 8% with or without the vitamins,[23] and a 2025 sprint-training trial again found a blunted muscle response but the same gain in cycling performance.[24] In one trial in older men, though, the vitamins shrank the gain in muscle mass.[25] They bring no training benefit, so there is no reason to take them. Antioxidants in food, at food doses, are a different matter; the problem is the concentrated pill taken alongside the very stressor it cancels.
So the most effective mitochondrial interventions are not pills but hormetic stressors. Exercise has the best human evidence; fasting, heat and cold impose the same kind of transient, survivable stress, though their mitochondrial effects in people are much less well tested.
What actually helps — the hormetic levers
Moderate for exercise; Weak for fasting, heat, cold and sleep. For cold, and for fasting's effect on mitochondrial cleanup, the evidence is largely animal and the few human fasting studies found little; for heat and sleep it comes from a few small, short human trials. None of the five has been tested against lifespan.
Exercise is the most potent lever, and both intensities work. Steady, conversational-pace aerobic work — zone 2 training — and harder intervals both build more mitochondria, and by about the same amount: across hundreds of human training studies, mitochondrial content rose by roughly a quarter either way. What predicts the gain is total training load — how much you train times how hard — so intervals get there in less time, while easy work is the intensity most people can do a lot of.[26] The response does not fade with age: in the same analysis the gains were not influenced by age, sex or disease, and the least fit gained the most. Strength training also improves mitochondrial function in older adults, alongside its role against muscle loss, though whether it also builds more mitochondria in older adults, as it does in the young, is inconsistent across studies.[27] The dose still matters, as the hormesis logic predicts: in a four-week experiment in healthy volunteers, steadily piling on more training led, after the heaviest week, to a sharp drop in mitochondrial function and worse blood-sugar control.[28] Mixing mostly easy sessions with a smaller dose of hard ones is a sustainable way to build training load, though that mix has not been shown to be a stronger mitochondrial stimulus in itself. The gains show up as a higher VO₂ max, one of the strongest predictors of mortality.
Fasting and caloric restriction may switch on the cleanup crew — so far shown mainly in animals. When energy runs low, the cell's low-fuel sensor activates and drives mitophagy — the selective clearing of damaged mitochondria — and fasting also dampens the growth pathway mTOR (the mechanistic target of rapamycin).[29] In people it is thin: in a six-month trial in 36 overweight adults, cutting calories by a quarter raised muscle mitochondrial DNA but not mitochondrial enzyme activity,[30] and in healthy men an 8-hour fast did not switch on the cleanup markers in muscle.[31] Nor does meal timing add a metabolic edge of its own: trials that match calories find time-restricted eating and fasting do about as well as ordinary calorie cutting, not better.[32][33] The animal machinery is the same as that described under deregulated nutrient sensing.
Heat and cold both qualify as hormesis. Heat triggers the heat-shock response, which makes protective chaperone proteins. In small human trials, two hours a day of deep heating of the thigh muscle raised heat-shock proteins and the muscle's mitochondrial capacity,[34] and during 10 days of leg immobilisation it cut the loss of mitochondrial function from about 27% to about 8%.[35] That was targeted muscle heating, not sauna: whether ordinary sauna sessions do the same is untested, and whether this explains sauna's observational mortality association is unproven. Cold exposure is a different stimulus: in mice it raises the activity of mitochondria-building genes in muscle[36] — but whether brief cold-plunge protocols do anything similar in humans is unsettled, and most of the human metabolic evidence comes from prolonged mild-cold air exposure rather than ice baths.
Sleep and circadian alignment protect the machinery. Mitochondrial energy output and antioxidant defences are synchronised to the 24-hour clock.[37] In flies and rodents, sleep loss disturbs mitochondrial gene expression and raises oxidative stress in the brain.[38] In people the evidence is small but direct: in a trial of 24 healthy young men, five nights limited to 4 hours in bed cut muscle mitochondrial function and worsened blood-sugar handling, while in the group that also did three high-intensity interval sessions that week, neither happened.[39] Animal studies also suggest mitochondria can make their own melatonin and use it as a local antioxidant, but this has not been shown in living human tissue, and that supply appears not to depend on the night-time melatonin the pineal gland releases.[40] So the case for consistent, dark-aligned sleep rests on the sleep-loss evidence, not on melatonin.
The supplement landscape
Weak. Every compound here is mechanism-rich and outcome-poor. None has shown a benefit on a hard outcome such as disease or survival; the few randomised trials that met their main goal were small, short and measured stand-in markers, and the larger muscle and performance trials mostly missed theirs. The site covers the pharmacology in depth under geroprotectors.
- Urolithin A has the most human data of the group, and it is more equivocal than its marketing. It is a postbiotic — a compound gut bacteria make from pomegranate and berry precursors, though roughly 1 in 10 people make none at all, and the amount everyone else makes varies widely[41][42] — and it directly stimulates mitophagy. The two muscle trials pull in different directions. ATLAS, in 88 overweight middle-aged adults over four months, found hamstring strength rose about 10–12% on urolithin A while it fell about 10% on placebo (front-of-thigh strength did not change significantly); C-reactive protein (a general marker of inflammation) fell, significantly only at the higher dose, and peak oxygen uptake rose about 10% in that group but fell just short of significance against placebo — but the trial missed its primary endpoint of peak power output, and the trial was sponsored by the company that makes the product, with all but one author its employees, board members or advisers.[43] A second trial in 66 adults aged 65–90 missed both of its primary endpoints — no improvement in six-minute walk distance or in maximal ATP production — and the muscle-endurance signal it did show at two months had gone by four months as the placebo group caught up.[44] The two results have not been reconciled. Two later four-week trials in middle-aged adults split the same way: one, funded by the maker, in 50 healthy adults met its main goal of shifting immune cells toward a younger-looking profile,[45] while an independent one in 50 adults with obesity found no improvement in artery function.[46] The honest reading is a small, mostly maker-funded set of trials with no consistent benefit — a strength signal in middle-aged adults and an immune-marker shift, not established benefit in older people.
- NAD⁺ precursors — nicotinamide mononucleotide and nicotinamide riboside — reliably raise blood levels of NAD⁺, a coenzyme essential to energy production and repair that declines with age. The repletion is real; the functional payoff in muscle is where the evidence is weakest. A randomised trial gave 32 adults aged 55–80 nicotinamide riboside with pterostilbene around an induced muscle injury and found no improvement in muscle stem-cell recruitment or in any measure of muscle recovery and regeneration.[47] In the other direction, a 10-week trial in 25 postmenopausal women with overweight or obesity and prediabetes found that 250 mg a day of nicotinamide mononucleotide improved how well their muscle responded to insulin — its main goal — but blood sugar, body composition, blood pressure and the muscle's mitochondrial capacity did not change.[48] See deregulated nutrient sensing for what raising NAD⁺ does and doesn't buy elsewhere.
- MitoQ, an antioxidant built to collect inside mitochondria, raised a standard measure of artery function (flow-mediated dilation) by about 40% against placebo in a 6-week crossover trial of 20 healthy adults aged 60–79 whose artery function was already below normal.[49] That is promising, but the trial was small and short, and it measured a stand-in rather than heart attacks or strokes; one of the compound's developers was a co-author.
- Coenzyme Q10 is part of the energy chain itself. In rare inherited CoQ10 deficiency, early high-dose treatment can help, with the response depending on the gene involved. For statin muscle aches the evidence is contested: pooled small trials show at most a slight benefit,[50] and a placebo produces much of the same complaint (see Coenzyme Q10). For healthy adults the longevity case is thin.
- Rapamycin blocks mTOR, the growth pathway described above, and is among the most studied candidate geroprotectors. Researchers hope low, intermittent doses can keep its benefits while avoiding its metabolic and immune side effects, but the right dose and schedule in people are still unknown.[51] Its placebo-controlled trials in healthy adults have so far shown no benefit: PEARL missed its main goal,[52] and in a 2026 trial in 40 sedentary people aged 65–85, weekly rapamycin did not improve the gains from an exercise programme — if anything the placebo group did slightly better, with fewer side effects[53] (geroprotectors). Firmly investigational, not a routine intervention.
- Elamipretide, a drug that binds a fat molecule in the inner mitochondrial membrane, received US accelerated approval in 2025 to improve muscle strength in Barth syndrome, an ultra-rare inherited disease.[54] It says nothing about normal aging.
The honest ranking: urolithin A has a mixed set of small, mostly maker-funded trials — one immune-marker win, no consistent benefit; the NAD⁺ precursors raise a meaningful molecule, with one small trial showing better muscle insulin sensitivity and none yet showing a functional gain in muscle; MitoQ has one small positive trial on artery function; and the rest is mechanism in search of outcomes. All of it sits well behind the hormetic levers above.
A caution on mouse data: the thermoneutrality gap
Methodological caveat, not a harm signal: a standing reason to discount striking mitochondrial metabolic results in mice — but not a blanket dismissal of mouse work, as the counterweight below shows.
One reason to be sceptical of dramatic mitochondrial claims is that most of the underlying biology comes from laboratory mice — and standard mouse housing quietly distorts exactly this system. Labs keep mice at around 22 °C for human comfort, but that is well below a mouse's comfort zone (about 30 °C), so caged mice live under constant mild cold stress: at 22 °C a mouse spends more energy making heat than its whole body uses at rest,[55] pushing its total energy use to roughly three times its resting rate.[56] Because cold is itself a mitochondrial stimulus, standard housing chronically revs their metabolism in ways that don't match a thermally comfortable human — shifting baseline glucose handling and brown-fat activity. The lifespan findings are sturdier than the metabolic ones, though: in one long-lived mutant strain (mice lacking the growth-hormone receptor), housing them at 30 °C from weaning cut their heat production by about a quarter but did not shorten their lives — females even lived 14% longer — so the gap is a reason to discount specific metabolic readouts, not to throw out the lifespan findings.[57] It is a concrete example of why a striking mitochondrial result in cold-stressed mice may not carry over to people, and why the human-trial evidence is what deserves the weight.
What this does and doesn't tell you
What it tells you: mitochondrial dysfunction is a real, integrative driver of aging — it links energy decline to inflammation (through the cGAS-STING DNA alarm) and, in early work, to post-infectious chronic fatigue syndrome, and, on observational evidence only, it tracks with frailty and dementia through the muscle–brain axis. Most importantly, mitohormesis explains why exercise works: it is a controlled dose of mitochondrial stress that triggers renewal, and fasting, heat and cold are thought to act the same way, though for them the human mitochondrial evidence is still thin. This is one of the few hallmarks where the human intervention evidence is more than mechanism — though still on surrogate measures, never on lifespan.
What it doesn't tell you: that any supplement extends human life by "boosting" mitochondria — none has shown a benefit on a hard outcome, and the few small trials that met their main goal measured stand-in markers such as insulin sensitivity, artery function or immune-cell profiles. Nor that more antioxidants are better; taken in large isolated doses around training they can blunt the adaptation. Nor that the dramatic mouse results will translate. The durable message is the same one that runs through this site: the proven way to keep your mitochondria young is to use them, rest them, and stress them in the right doses.
Further reading
- Srivastava S. The mitochondrial basis of aging and age-related disorders. Genes (Basel) 2017.[58]
- Wei P, et al. Mitochondrial dysfunction and aging: multidimensional mechanisms and therapeutic strategies. Biogerontology 2025.[59]
- Gorgori-Gonzalez A, et al. Leveraging mitochondrial stress to improve healthy aging. Sports Med Health Sci 2026.[60]
- Ding W, et al. Mitochondrial DNA leakage triggers inflammation in age-related cardiovascular diseases. Front Cell Dev Biol 2024.[61]
- Salminen A, Kaarniranta K, Kauppinen A. Activation of cGAS-STING signaling in senescent cells promotes the aging process by remodeling the functions of the immune system. Biogerontology 2025.[62]
- Tian Q, et al. Skeletal muscle mitochondrial function predicts cognitive impairment and is associated with biomarkers of Alzheimer's disease and neurodegeneration. Alzheimers Dement 2023 — Baltimore Longitudinal Study of Aging.[63]
- Wang P, et al. WASF3 disrupts mitochondrial respiration and may mediate exercise intolerance in myalgic encephalomyelitis/chronic fatigue syndrome. Proc Natl Acad Sci U S A 2023.[64]
- Singh A, et al. Urolithin A improves muscle strength, exercise performance, and biomarkers of mitochondrial health in a randomized trial in middle-aged adults. Cell Rep Med 2022 — ATLAS; missed its primary endpoint; 8 of 9 authors tied to the maker.[65]
- Liu S, et al. Effect of urolithin A supplementation on muscle endurance and mitochondrial health in older adults. JAMA Netw Open 2022 — missed both primary endpoints.[66]
- Jensen JB, et al. A randomized placebo-controlled trial of nicotinamide riboside and pterostilbene supplementation in experimental muscle injury in elderly individuals. JCI Insight 2022.[67]
- James CM, et al. How murine models of human disease and immunity are influenced by housing temperature and mild thermal stress. Temperature (Austin) 2023.[68]