Name: Mitochondrial Health: The Guaranteed Lever Is Training, Not the Stack
Creator: Realised
Published: 2026-07-01

Summary

Mitochondrial health is real and worth caring about, but the honest map runs by evidence strength rather than by marketing spend: aerobic and resistance training is the one guaranteed lever that reproducibly builds mitochondria; the free hormetic levers (fasting, cold, heat, sleep) are plausible; and almost the entire paid "mitochondrial support" tier (NAD precursors, resveratrol, spermidine, urolithin A, PQQ, MitoQ, methylene blue) moves biomarkers in humans while leaving healthspan outcomes unproven, so both the biohacker claim that a stack is the anti-aging key and the dismissive claim that

Why Strong

Tier: the hub carries a HIGH-confidence organising thesis, with per-lever tiers spanning Foundational to Experimental. The load-bearing claim of the entry, that training is the guaranteed lever, hormesis is the plausible free tier, and the paid tier is mostly unproven, is itself Foundational, resting on the strongest human and mechanistic evidence in the field (exercise biogenesis, mitohormesis, human caloric-restriction biogenesis, and a well-documented biomarker-vs-outcome gap for the supplements).

NOT a lower tier for the headline, because the ranking is not speculative, it is exactly what the evidence hierarchy shows; the uncertainty lives inside individual levers, not in the ordering.

NOT "Foundational" as a blanket label, because a hub that pointed at nothing but settled facts would be misleading, several of the levers it routes to are Emerging or Experimental, and the entry's honesty depends on preserving that spread rather than averaging it away.

The per-sub-area tier split (read this, not just the headline):
• Mitochondria as ATP/ROS-signal/apoptosis hub; mitohormesis; exercise biogenesis: Foundational/Strong.
• Caloric restriction / fasting biogenesis (human): Moderate; human mitophagy: Emerging.
• Cold / heat / sleep as mitochondrial levers: Moderate/Emerging (mechanism-led).
• CoQ10 in the statin/deficiency niche: Moderate and contested; creatine, muscle: Strong; creatine, cognition: Moderate/Emerging.
• NAD precursors, resveratrol, spermidine, urolithin A, MitoQ, PQQ: Emerging to Experimental (biomarkers move, outcomes unproven).
• Methylene blue: Experimental, plus a real safety flag.

Practical takeaway

This is a hub. The section below is the tiered lever map: it routes, it does not re-argue. For each lever, hold the tier and go to the spoke for the detail.

GUARANTEED (Foundational/Strong, free) — do these first.
• Aerobic base training. Reproducibly builds mitochondrial density and function via PGC-1α; the single most robust lever. Go to physical_zone2_cardio for prescription.
• High-intensity / VO2max work and its longevity link. Strong per-minute mitochondrial stimulus and the fitness metric most tied to mortality; pairs with aerobic base rather than replacing it. Go to vo2max_and_longevity.
• Resistance training (for muscle mitochondrial content specifically) belongs here too; it is covered under the physical pillar's strength entries rather than a dedicated mitochondrial spoke.

PLAUSIBLE-HORMETIC (Moderate/Emerging, free) — worthwhile, don't over-claim.
• Fasting / caloric restriction. Genuine human biogenesis data (~35% muscle mtDNA over six months) and a strong animal mitophagy trigger; human mitophagy is promising-not-proven. Go to diet_extended_fasting.
• Cold exposure. Plausible biogenesis/brown-fat/uncoupling stimulus; human mitochondrial-endpoint data are thinner. Go to deliberate_cold_exposure_stress.
• Heat / sauna. Heat-shock-protein and mitochondrial-support mechanisms, mechanistically plausible, human mitochondrial data lighter. Go to sauna_heat_exposure.
• Sleep is a plausible hormetic support with no dedicated mitochondrial spoke; treat it as foundational recovery infrastructure, not a mitochondrial "intervention."

TARGETED-SUPPLEMENT (Moderate, paid) — real in a niche, not for everyone.
• CoQ10 / ubiquinol. Defensible for statin-associated muscle symptoms, documented deficiency, and primary mitochondrial disease, and even there the RCTs are mixed; no good case in healthy, non-deficient people. Go to coq10_evidence_and_use_cases.
• Creatine. A real phosphocreatine energy buffer, strong for muscle, modest-emerging for cognition, and explicitly not a biogenesis agent. Go to creatine_comprehensive.

SPECULATIVE (Emerging/Experimental, paid) — biomarkers move, outcomes unproven.
• NAD precursors (NMN, NR). Reliably raise blood NAD+; no proven human healthspan outcome. Go to nad_precursors_nmn_nr_longevity.
• Resveratrol. Sirtuin-pathway story, poor bioavailability, mostly-null human data. Go to resveratrol_and_longevity_claims.
• Spermidine. Autophagy/mitophagy story, mostly observational/small human data. Go to spermidine_longevity_evidence.
• No KB spoke yet (discussed in prose only): urolithin A (Mitopure) is the most human-supported of this cluster and still only modest (flagship trial missed its primary endpoint); MitoQ rests on a single small vascular RCT; PQQ is thin. Do not present any of these as established mitochondrial-health interventions. Methylene blue is NO LONGER prose-only — it now has its own spoke, methylene_blue_cognition_and_mitochondria (Experimental), which owns the MAOI/serotonin-syndrome safety wall, the biphasic dose reversal, the G6PD contraindication and the failed phase-3 tau programme. Route every methylene-blue query there rather than answering from this hub.

The framing to hold: if someone wants to improve their mitochondria, the answer is training, then the free hormetic levers, and only then, in the right medical niche, a specific supplement. The order is by evidence, not by price. When it's a signal, not a lever: persistent exercise intolerance, disproportionate fatigue with muscle weakness, or exertional symptoms out of proportion to effort are not a cue to buy a mitochondrial stack, they route to clinical assessment (primary mitochondrial disease, cardiac, thyroid, and anaemia causes belong to a clinician).

Evidence detail

Why This Entry Exists

"Mitochondrial health" is one of the most abused phrases in the longevity market. It is used to sell a premium tier of supplements on the strength of mouse lifespan data and human biomarker shifts, and it is used, in the opposite direction, as a punchline by people who have decided the whole topic is pseudoscience. Both are wrong, and they are wrong for the same reason: they ignore the actual shape of the evidence, which is lopsided. One lever is rock-solid and free, a few are plausible and free, and the expensive ones are mostly promissory.

This entry is a hub. Its job is not to re-argue any single lever, but to organise them into an honest tier map and route each to the spoke that carries its detail. The naive framing that dies here is the old free-radical theory, that reactive oxygen species (ROS) are pure damage and antioxidants are the fix. The mainstream reframe is mitohormesis: low-dose, exercise-generated ROS is an adaptive signal, and blunting it with high-dose antioxidants can block the very adaptations you were training for. That single reversal explains most of why the "load up on antioxidants for your mitochondria" story is backwards, and why the guaranteed lever turns out to be the one nobody profits from.

What bad advice this protects against, in all directions:
• "Take this NAD/CoQ10/urolithin stack, it's the key to anti-aging." The paid tier reliably moves biomarkers (blood NAD+ rises, mitochondrial markers shift) but no human trial has shown a hard healthspan or longevity outcome. This is the textbook biomarker-vs-outcome gap dressed up as a breakthrough.
• "Mitochondrial health is just supplement woo, ignore it." Equally wrong, and it cedes the strongest true point in the field: exercise centrally and reproducibly improves mitochondrial function and density. The biology is not woo; the supplements are the weak part.
• "Megadose antioxidants (vitamin C and E) to protect your mitochondria from exercise damage." The direct human evidence runs the other way: high-dose antioxidants around training can prevent the metabolic adaptations you exercised to get.
• "Fasting/autophagy clears out your old mitochondria, the mouse data prove it." Caloric restriction has genuine human biogenesis data, but human mitophagy (the turnover half) is measured poorly and largely inferred from animals. Importing mouse strength into a human claim is the standard overreach here.
• "CoQ10 boosts everyone's energy." Legitimate in a narrow medical niche (statin myalgia, deficiency, primary mitochondrial disease) and even there the RCTs are mixed; in healthy people with no deficiency there is no good evidence it does anything.

This entry owns the mitochondrial-health overview and the tiered lever map (guaranteed → plausible-hormetic → targeted-supplement → speculative) and routes each lever to its spoke. It defers every lever's evidence detail to that spoke and does not re-litigate it here.

Evidence

Read the tiers, not the thesis. The claims below are ordered by how strongly the science supports them, not by how loudly they are sold.

Mitochondria as the metabolic hub, and the death of the naive free-radical theory (Foundational).

1. Mitochondria generate most cellular ATP via oxidative phosphorylation, but also emit ROS as signals, gate apoptosis, and buffer calcium; ROS are not simply damage. Consensus biochemistry describes the electron transport chain (Complexes I–V) making ATP, cytochrome-c release gating intrinsic apoptosis, and low-dose ROS acting as second messengers that drive adaptation, the mitohormesis reframe. Age-related decline is better described as a mix of falling mitochondrial density, reduced per-mitochondrion function, and, most consistently, impaired quality control (mitophagy/turnover) than as simple accumulated free-radical damage; the relative contribution varies by tissue and is an active research area. (Foundational consensus across many sources, anchored empirically by the mitohormesis literature; stated generically to avoid welding a multi-source consensus to one paper. No commercial interest on the consensus side, whereas the older "antioxidants fix aging" framing is what sells antioxidant pills.)

Mitohormesis: the human proof that exercise ROS is a required signal (Strong Evidence).

2. High-dose antioxidant supplementation can prevent the health-promoting metabolic adaptations to exercise. In a four-week study, vitamin C (1000 mg/day) plus vitamin E (400 IU/day) abolished exercise-induced improvements in insulin sensitivity and blunted induction of endogenous antioxidant defenses and of PGC-1α, in a small cohort of healthy young men, versus exercise alone. This is a load-bearing mechanistic signal, not a settled universal law: the sample was small (n=39), young, healthy, and the endpoint was a surrogate (insulin sensitivity), and the broader interference literature is dose- and context-dependent and mixed; food-level antioxidants are not implicated, only high-dose isolated supplements around training. (Ristow M, Zarse K, Oberbach A, et al. PNAS 2009;106(21):8665–8670. Independent academic work; it directly contradicts antioxidant-vendor claims, which is precisely why some vendors pivoted to selling "mito-targeted" antioxidants instead.)

The guaranteed lever: exercise drives mitochondrial biogenesis (Foundational).

3. Aerobic training (and resistance training for muscle mitochondria) is by far the most robust driver of mitochondrial biogenesis, acting largely through the PGC-1α program. Exercise triggers a rapid, transient surge in PGC-1α transcription in human skeletal muscle and drives measurable biogenesis, the single most reproducible intervention in the field. The honest nuance: PGC-1α is the master coactivator but is not strictly required, knockout mice still mount a biogenic response to endurance exercise via redundant pathways, so "exercise builds mitochondria" is Foundational while "PGC-1α is THE one required switch" is an overstatement. Precise zone-2-vs-HIIT rankings are less settled than the umbrella claim and should not be asserted as proven. (Pilegaard H, Saltin B, Neufer PD. J Physiol 2003;546(Pt 3):851–858; redundancy nuance, Rowe GC et al. PLoS One 2012;7(7):e41817. Independent academic; there is nothing to sell on the exercise side, which is partly why it is under-marketed relative to pills.)

The free hormetic levers: caloric restriction / fasting, and plausibly cold, heat, sleep (Moderate).

4. Caloric restriction increases human muscle mitochondrial biogenesis; fasting is a strong mitophagy trigger, but human mitophagy data are limited. Six months of caloric restriction increased muscle mitochondrial DNA content by roughly 35% (about 21% with CR plus exercise) alongside upregulated PPARGC1A/SIRT1 and reduced DNA damage in overweight non-obese adults, so CR→biogenesis in humans is genuinely supported. Mitophagy specifically is strong in animal models and promising but not proven in humans, because in-vivo human mitophagy markers are few and heterogeneous. Cold exposure, heat/sauna, and sleep are mechanistically plausible and free but rest more on mechanism and animal data, keep them at Moderate/Emerging, not guaranteed. (Civitarese AE, Carling S, Heilbronn LK, et al. PLoS Medicine 2007;4(3):e76; autophagy caveat, Bagherniya M, Butler AE, Barreto GE, Sahebkar A, review, Ageing Res Rev 2018;47:183–197 (mitophagy is the mitochondrial subset of the autophagy it reviews). Independent academic; these levers are free, so no vendor pushes them, which supports the "you cannot buy mitochondrial health" thesis. The fasting/autophagy market overstates the human mitophagy case by importing mouse strength.)

Targeted supplements: real in a niche, oversold outside it (Moderate).

5. CoQ10/ubiquinol is defensible for statin-associated muscle symptoms, deficiency, and primary mitochondrial disease, but the RCT evidence is genuinely mixed, and it is oversold for healthy people. CoQ10 is a real electron-transport-chain carrier, and statins lower it, a plausible mechanism for statin myalgia. Some meta-analyses report symptom benefit while several rigorous RCTs and other analyses find no significant effect on muscle symptoms or creatine kinase, so treat it as promising-but-contested, not settled, and note the large nocebo effect in statin-myalgia trials. For healthy people with no deficiency there is no good evidence CoQ10 boosts energy or longevity. (Positive pole: Qu H, et al. J Am Heart Assoc 2018;7(19):e009835, 12 RCTs, 575 patients, symptom improvement; multiple RCTs/meta-analyses report null effects. The supplement industry generalises a plausible medical niche into "everyone needs CoQ10 for energy"; the honest reading is a legitimate targeted use, oversold for the general population.)

6. Creatine is a real phosphocreatine energy buffer with strong muscle benefit and modest emerging cognitive benefit, but it is not a biogenesis agent. Creatine's muscle-performance and lean-mass benefit is well established. The mitochondrial-adjacent rationale is genuine: creatine/phosphocreatine buffers ATP under high demand. Cognition meta-analyses show a small but significant memory benefit (standardized mean difference +0.29, 95% CI 0.04–0.53), strongest in older adults (the young-adult subgroup was null), so "some cognitive benefit" is fair at Moderate/Emerging strength, with mechanism uncertainty (at least one analysis found no change in brain phosphocreatine). Do not conflate an energy buffer with building mitochondria. (Muscle: broad consensus, ISSN position stand. Cognition: Prokopidis K, et al. Nutrition Reviews 2023;81(4):416–427, memory SMD +0.29 (95% CI 0.04–0.53). Creatine is cheap and off-patent, so hype is comparatively low relative to evidence, one of the few supplements where evidence exceeds marketing.)

The speculative/hyped tier: biomarkers move, human outcomes do not (Emerging → Experimental).

7. NAD precursors (NMN, NR) reliably raise blood NAD+ but have no proven human healthspan or longevity outcome. NR (and NMN) dose-dependently and safely raise blood NAD+, roughly 22% / 51% / 142% at 100 / 300 / 1000 mg in a controlled trial, but human RCTs are dominated by safety and biomarker endpoints; the dramatic mouse benefits (glucose, weight, lifespan) have not translated into demonstrated human disease or healthspan outcomes, and early trials in mild cognitive impairment, Parkinson's, and long-COVID have not established efficacy. This is the cleanest biomarker-vs-outcome gap in the field. (Conze D, Brenner C, Kruger CL. Sci Rep 2019;9:9772, dose-response/safety; Martens CR et al. Nat Commun 2018, NAD/blood-pressure signal. Publication and commercial bias are heavy, several trials are manufacturer-funded; the industry's incentive is to equate "raises NAD+" with "anti-aging," which is exactly the overclaim.)

8. Resveratrol and spermidine act on plausible longevity pathways but have weak, inconsistent human translation. Both extend lifespan in some invertebrate/rodent models (resveratrol→sirtuins/SIRT1; spermidine→autophagy/mitophagy). Human data are weak: resveratrol has poor bioavailability and mostly-null human metabolic RCTs; spermidine has associational and small-trial human data (some memory signals) but no established healthspan outcome. Mechanism-rich, human-poor. (Characterised as preclinical-dominant; no human outcome trial establishes a healthspan endpoint, so stated generically rather than welded to a false citation. Positive human resveratrol data are inconsistent and often industry-linked; the industry markets both as "sirtuin/autophagy activators for longevity" on animal data.)

9. Prose-only specialty tier, no KB spoke yet: urolithin A, MitoQ, PQQ, methylene blue range from modest-real to preclinical-and-risky. Urolithin A (Mitopure) is a mitophagy activator whose flagship middle-aged-adult RCT reported ~12% muscle-strength gains and mitochondrial-biomarker improvements but missed its primary endpoint (peak power), real-but-modest, not decisive. MitoQ (mito-targeted CoQ) rests largely on one small RCT (n=20, older adults) improving vascular endothelial function over six weeks. PQQ human data are thin. Methylene blue is an electron cycler with preclinical interest but a genuine safety hazard, it is an MAO inhibitor with serotonin-syndrome risk alongside serotonergic drugs, and must not be presented as a benign nootropic. (Urolithin A: Singh A, et al. Cell Reports Medicine 2022;3(5):100633, primary endpoint not met. MitoQ: Rossman MJ, et al. Hypertension 2018;71(6):1056–1063, n=20. Strong commercial push frames these as the premium anti-aging mitochondrial tier on modest/preclinical data; methylene blue's drug-interaction risk is real and downplayed by vendors.)

The hub-level integration (Foundational, as a judgment).

10. Exercise dwarfs everything; hormesis is plausible and free; supplements are mostly marketing with a few targeted medical exceptions. The tier structure above is the argument: the only Foundational/Strong lever is training; the free hormetic levers are Moderate; the paid levers split into a few legitimate targeted uses (CoQ10 in the statin/deficiency niche, contested even there; creatine as a real energy buffer) versus a large hyped tier where biomarkers move but human healthspan outcomes are unproven. (Synthesis of the sources above; an evidence-weighted judgment, not a single trial. Legitimate medical exceptions, mitochondrial disease, statin myalgia, documented deficiency, must not be flattened into "all supplements are useless.")

Mechanism

Why mitochondria sit at the metabolic centre. They run oxidative phosphorylation to make most cellular ATP, and in doing so they read the cell's fuel and redox state. They also decide cell fate (intrinsic apoptosis via cytochrome-c release) and buffer calcium. Crucially, the ROS they emit are not only damage, they are signals: a controlled ROS pulse tells the cell to adapt.

Why exercise is the master switch (mitohormesis). Contracting muscle raises calcium, AMP/ATP ratio, and ROS, all of which converge on the PGC-1α transcriptional program that drives mitochondrial biogenesis and capillarisation. This is the mechanistic reason the guaranteed lever is training. It is also why blunting the ROS pulse with high-dose antioxidants can block the adaptation, you are silencing the signal, not protecting the machinery. PGC-1α is the coordinator but not the sole gate; redundant pathways can carry the response, so the program is robust rather than fragile.

Why fasting and caloric restriction plug into the same circuitry. Energy scarcity raises AMPK and NAD+-dependent sirtuin signalling, which feed the same biogenesis program and, in animals, strongly induce mitophagy, the clearing of damaged mitochondria. Human biogenesis under caloric restriction is measurable; human mitophagy is the harder half to prove because we lack good in-vivo markers, so the mechanism is confident and the human turnover data are thin.

Why the paid levers are mechanistically seductive but clinically unproven. Each has a real target: NAD precursors raise a genuinely important coenzyme; resveratrol nudges sirtuins; spermidine and urolithin A push autophagy/mitophagy; CoQ10 is a real chain carrier; creatine is a real ATP buffer. A plausible mechanism is necessary but not sufficient, the recurring failure is that moving the mechanistic marker (blood NAD+, a biomarker panel) has not translated into a demonstrated human outcome. Mechanism is where these stories start; outcomes are where most of them stop.

Risks And Contraindications

• Do not over-claim the paid tier. Raised blood NAD+, a shifted biomarker panel, or a mouse lifespan result is not a demonstrated human benefit. Never present a biomarker move as an outcome.
• High-dose isolated antioxidants around training can backfire. Megadose vitamin C/E near workouts may blunt the adaptations you trained for; this does not apply to food-level intake.
• CoQ10 is a targeted, contested use, not a general energiser. Reserve the claim for statin myalgia, deficiency, or primary mitochondrial disease, and even there present the RCT evidence as mixed and note the large nocebo effect in statin trials.
• Methylene blue is a drug, not a supplement. As an MAO inhibitor it carries serotonin-syndrome risk with SSRIs/SNRIs and other serotonergic agents; do not surface it as a benign nootropic.
• Fasting/caloric restriction is not for everyone. Contraindicated or requiring care in eating-disorder history, pregnancy, being underweight, type 1 diabetes, and certain medication regimens, defer the specifics to diet_extended_fasting.
• Do not flatten legitimate medical exceptions. Primary mitochondrial disease, documented deficiency, and statin myalgia are real clinical situations where targeted supplementation is appropriate under a clinician; "supplements are mostly hype" is a population statement, not medical advice for these cases.
• Symptoms of true mitochondrial or muscle disease route to care, not to a stack. Disproportionate exertional fatigue with weakness is a clinical signal.

Controversy

Nature: the controversy is not about whether mitochondria matter, that is settled, but about how much you can buy. A large longevity-supplement industry pulls the frame toward "mitochondrial support stacks are the anti-aging key," while a dismissive counter-camp pulls it toward "it's all supplement woo." The honest position sits between and annoys both.

Position A, the pro-supplement/biohacker take. Best evidence: several compounds have real mechanisms and move real biomarkers, NAD precursors robustly raise blood NAD+, urolithin A improved strength and mitochondrial markers in a randomised trial, creatine and CoQ10 have genuine physiology behind them. Where it overreaches: it equates mechanism and biomarker movement with proven healthspan benefit, leans on mouse and worm lifespan data, and quietly ignores that the flagship human trials are small, short, often manufacturer-linked, and frequently miss their primary endpoints.

Position B, the dismissive take. Best evidence: the paid tier really is oversold, the outcome data really are missing, and the marketing really does outrun the science. Where it overreaches: it throws out the strongest true finding in the field, that exercise reproducibly and centrally improves mitochondrial function, and it dismisses legitimate targeted medical uses (mitochondrial disease, statin myalgia, deficiency) along with the hype.

The funding/bias dimension, cui bono, both ways. The heaviest commercial and funding bias sits on the paid-supplement side: the entire "buy the stack" revenue model depends on equating biomarkers with benefit, and several key trials are manufacturer-funded. But bias is not only commercial, the dismissive camp has its own incentive to sound rigorous by rejecting everything, which conveniently lets it ignore the exercise evidence it would otherwise have to act on. The clean signal is the one with no product attached: training. Nobody sells it, and it has the strongest data.

Realised Position: Mitochondrial health is real, and the guaranteed lever is training, full stop. The free hormetic levers are worth doing and honestly labelled as plausible. The paid tier is mostly marketing with a few genuine, narrow medical exceptions, and its central overclaim, biomarker equals benefit, should be named every time. Both slogans are wrong: you cannot buy mitochondrial health, and it is not woo. Route people to the lever that matches the evidence, in evidence order.

Cross-Pillar Connections

Genuinely cross-pillar: mitochondrial function is where the physical, diet, and recovery pillars converge, so this hub routes across all of them.
• physical_zone2_cardio owns aerobic-base prescription and the zone-2 debate; this entry defers the "how" and holds only that aerobic training is the guaranteed biogenesis lever.
• vo2max_and_longevity owns the fitness-mortality link and high-intensity prescription; this entry defers the detail and holds the routing.
• deliberate_cold_exposure_stress owns cold-exposure protocol and its full evidence; this entry holds only the plausible-hormetic label.
• sauna_heat_exposure owns heat/sauna dosing and outcomes; this entry holds only the plausible-hormetic label.
• diet_extended_fasting owns fasting/caloric-restriction protocol, safety, and contraindications; this entry holds the biogenesis-yes / human-mitophagy-thin framing and defers the rest.
• coq10_evidence_and_use_cases owns the CoQ10 evidence, dosing, and statin-niche detail; this entry holds only the "targeted, contested, not for healthy people" placement.
• creatine_comprehensive owns creatine's full evidence base; this entry holds only that it is an energy buffer, not a biogenesis agent.
• nad_precursors_nmn_nr_longevity owns the NAD-precursor evidence; this entry holds only the biomarker-vs-outcome-gap placement.
• resveratrol_and_longevity_claims owns resveratrol's evidence; this entry holds only its speculative placement.
• spermidine_longevity_evidence owns spermidine's evidence; this entry holds only its speculative placement.

What would change our mind

Falsifiability: explicit upgrade/downgrade criteria from source

• We would promote a paid lever out of "speculative" if a well-powered, independent (non-manufacturer-funded) human RCT showed a hard healthspan or disease outcome, not a biomarker, for NAD precursors, resveratrol, spermidine, or urolithin A.
• We would strengthen the CoQ10 statin claim if the meta-analytic picture converged on consistent, nocebo-controlled symptom benefit across high-quality trials, rather than reaching opposite conclusions.
• We would upgrade the human mitophagy story if validated in-vivo human mitophagy markers showed fasting/CR reproducibly clearing damaged mitochondria in people, closing the gap between animal and human data.
• We would revisit the exercise-supremacy framing only if a supplement demonstrated exercise-comparable mitochondrial biogenesis and function in humans, which nothing currently approaches.
• What would NOT move us: more mouse or worm lifespan data; another trial showing a supplement raises a biomarker; a bigger effect on a surrogate endpoint; or marketing that reframes "mito-targeted" delivery as if it resolved the missing-outcome problem. Mechanism and biomarkers are entry tickets, not conclusions.

Industry bias note

Structural incentives the evidence base may reflect

The commercial pressure in this field is unusually one-directional, and naming it is most of the entry's value.
• Toward the paid stack. The longevity-supplement industry's revenue depends on the reverse of this entry's thesis: that you can buy mitochondrial health. Its levers are the ones with heavy funding and publication bias, several NAD-precursor and urolithin-A trials are manufacturer-associated, and its rhetorical move is to equate a raised biomarker with a proven benefit. Watch for "mito-targeted" reframing (e.g. MitoQ) that repackages a criticised category rather than answering it.
• Toward dismissiveness, in the other camp. Less commercial but still a bias: rejecting the whole topic sounds rigorous and conveniently avoids the one demanding conclusion, that the effective lever is unglamorous training.
• The clean signal. The strongest lever, exercise, has no vendor, no sponsored trial, and no product page, and it has the best human data. That is not a coincidence, it is the tell. The honest synthesis sells nothing: train, use the free hormetic levers, and reserve supplements for the narrow medical cases where they are actually indicated.

Mitochondrial Health: The Guaranteed Lever Is Training, Not the Stack