Supplements · what the trials actually did
The supplements that work are the boring ones.
For women with a cupboard of supplements and doubts about most of them.
Graded on human trials alone: a mouse result cannot lift a grade, and a failed trial counts against it.
Sound familiar?
- Half the tubs in your cupboard were bought after a podcast, and you could not say which of them is doing anything.
- Someone told you magnesium for sleep, fish oil for your heart and NMN for everything else.
- You take vitamin D every winter and have never had your level tested.
Start here
What supplements can — and cannot — do
They fix deficiencies and nudge specific systems. Nothing in a capsule rivals training, sleep, diet, or sunscreen.
What supplements can — and cannot — do
They fix deficiencies and nudge specific systems. Nothing in a capsule rivals training, sleep, diet, or sunscreen.
Every supplement on this page competes with a brutal benchmark: the interventions we already know slow age-related decline. Resistance training, cardiorespiratory fitness, adequate protein, not smoking, and daily sunscreen each have evidence no pill approaches. Supplements earn a place in two situations: correcting a measured shortfall (vitamin D, magnesium, protein, fiber) or adding a specific, trial-tested effect (creatine for muscle, collagen peptides for skin).
The rest of the market — NAD+ boosters, senolytics, longevity blends — is running ahead of its evidence. Some of it is genuinely promising and worth watching. None of it has yet improved a human aging outcome (function, disease, lifespan) in a published randomized trial. This guide grades each option by what human data actually show, not by mechanism or mouse results.
A useful rule: the more a product promises (energy + skin + brain + immunity + longevity), the less any single claim is likely to survive scrutiny.
How we grade the evidence
Strong = repeated human trials with consistent outcomes. Limited = anecdote, animal data, or negative trials.
How we grade the evidence
Strong = repeated human trials with consistent outcomes. Limited = anecdote, animal data, or negative trials.
Each supplement below carries one of four tiers, based on human evidence only:
- Strong — multiple randomized controlled trials or meta-analyses with consistent, meaningful outcomes.
- Moderate — consistent human trials, but small, short, single-context, or industry-funded.
- Emerging — one or two small human trials, or reliable biomarker changes without outcome data yet.
- Limited — anecdote, animal-only data, marketing, or predominantly negative human trials.
Two honesty rules we apply throughout: spectacular mouse data cannot lift a tier (mice are not small humans — resveratrol taught the field that lesson), and a failed or null trial counts against a molecule even when its mechanism is beautiful.
Why it hasn't worked for you so far
Supplements: proven benefits or empty promises?
What's real: creatine with strength training, enough protein and fibre, collagen for skin, and vitamin D if a test shows you are low. What's not: blends without doses and most of the longevity aisle. Here's how to tell them apart:
-
The blend
No doses, no way to check
A proprietary blend need not say how much of each ingredient you get, and single-ingredient labels can be wrong too: NMN and urolithin A products tested in 2024 ranged from 28.6% above to 100% below their label claims. Buy one ingredient, at the trial dose, from a third-party-tested brand.
-
The longevity aisle
Sold on mice and biomarkers
Taurine extended mouse lifespan in 2023; by 2025, human data showed it does not even fall with age in most people. NAD+ boosters reliably raise NAD+ in the blood, yet a meta-analysis in older adults found no overall physical-function benefit. No NAD+ booster, senolytic or longevity blend has improved a human aging outcome in a published trial.
-
The one nobody mentions
A pill with nothing to fix
A supplement earns its place by correcting a measured shortfall or adding a trial-tested effect. In the 25,000-person VITAL trial, vitamin D given to people not selected for deficiency did not cut fractures (NEJM); creatine’s gains in muscle came on top of resistance training. Test before you correct, and train with what you take.
The science
Supplements — what the evidence says
Every option, strongest first. Open any row for the trials behind it.
The foundations
Strong evidence Creatine monohydrate
The best-evidenced supplement in this guide: with resistance training it adds measurable muscle and strength in older adults.
Creatine monohydrate
The best-evidenced supplement in this guide: with resistance training it adds measurable muscle and strength in older adults.
The best-evidenced supplement in this guide: with resistance training it adds measurable muscle and strength in older adults.
Creatine is the rare supplement with decades of consistent human data. Meta-analyses show that creatine plus resistance training produces greater gains in lean mass and strength than training alone, including in older adults — exactly the population losing 1–2% of muscle per year. Muscle is the closest thing aging research has to a retirement account: grip strength and lean mass predict mortality better than most blood tests.
Brain effects are earlier-stage: small trials suggest modest cognitive benefits under stress (sleep deprivation, vegetarian diets), but the aging-brain data remain preliminary — a bonus, not the reason to take it.
How to take it: 3–5 g/day of plain creatine monohydrate, every day, no loading phase needed. Expect ~1 kg of intracellular water weight in the first weeks — that is muscle hydration, not fat.
Safety: among the most studied supplements in existence; benign in healthy people. Reports of kidney harm at normal doses have not held up, but discuss it with a clinician if you have kidney disease.
- Best for
- Top pick: Thorne Creatine (NSF Certified for Sport)
Our pick
Thorne Creatine — creatine monohydrate and nothing else, 5 g a scoop, NSF Certified for Sport; 450 g.
Strong evidence Protein (and where whey fits)
Most adults over 50 under-eat protein. Hitting ~1.2 g/kg/day preserves the muscle that predicts how you age.
Protein (and where whey fits)
Most adults over 50 under-eat protein. Hitting ~1.2 g/kg/day preserves the muscle that predicts how you age.
Most adults over 50 under-eat protein. Hitting ~1.2 g/kg/day preserves the muscle that predicts how you age.
Total protein intake is the foundation the rest of this page stands on. A meta-analysis of trials in older adults supports protein supplementation combined with resistance training for lean mass and strength. European intake surveys consistently find adults over 50 clustering near the old minimum (0.8 g/kg) — set for avoiding deficiency, not for aging well.
Target: ~1.2–1.6 g/kg of body weight per day for adults focused on tissue maintenance, spread across meals of 25–40 g. Whey is the best-supported supplement form because of its leucine content; food-first works equally well if you actually hit the numbers.
Collagen peptides are a special case — a lower-quality protein for muscle, but with their own skin/joint evidence. They get a full guide of their own.
Strong evidence Fiber (psyllium) & fermented foods
Fiber has better human mortality data than any capsule in this guide: 15–30% lower all-cause mortality in the highest versus the lowest eaters, with the best risk reduction at 25–29 g/day.
Fiber (psyllium) & fermented foods
Fiber has better human mortality data than any capsule in this guide: 15–30% lower all-cause mortality in the highest versus the lowest eaters, with the best risk reduction at 25–29 g/day.
Fiber has better human mortality data than any capsule in this guide: 15–30% lower all-cause mortality in the highest versus the lowest eaters, with the best risk reduction at 25–29 g/day.
It is almost embarrassing that the best "longevity supplement" is fiber. The landmark Lancet series (185 prospective studies and 58 trials, ~135 million person-years) found 15–30% lower all-cause and cardiovascular mortality at the highest fiber intakes, with the best risk reduction at 25–29 g/day and a dose-response beyond. Psyllium is the pragmatic supplement form when food falls short, and modestly lowers LDL cholesterol as a bonus.
Fermented foods are earlier-stage but interesting: in a Stanford RCT, six daily servings increased microbiome diversity and lowered 19 inflammatory proteins over 10 weeks — a tiny trial (n=36), but cheap, safe, and food-first.
How to take it: count your food fiber first; add 5–10 g of psyllium daily with plenty of water if you're short. Ramp up slowly to spare your gut the surprise.
Moderate evidence Vitamin D — correct deficiency, skip "optimization"
Valuable if you are actually deficient (common in Northern Europe). Supplementing sufficient people showed no benefit in the big trials.
Vitamin D — correct deficiency, skip "optimization"
Valuable if you are actually deficient (common in Northern Europe). Supplementing sufficient people showed no benefit in the big trials.
Valuable if you are actually deficient (common in Northern Europe). Supplementing sufficient people showed no benefit in the big trials.
Vitamin D illustrates the difference between fixing a deficiency and chasing a number. In trials that did not select for deficiency, supplementation failed its endpoints — the 25,000-person VITAL trial found no reduction in cardiovascular events or cancer incidence, and its fracture analysis in NEJM was flatly null. Correcting genuine deficiency, by contrast, remains standard medicine — and deficiency is common at European latitudes, especially October to April.
How to take it: test first (a 25-OH-D blood test), then 800–2,000 IU/day of D3 if you're low. Retest once after 3–4 months.
Avoid megadoses: annual or monthly boluses have increased falls in trials, and chronic intakes above ~4,000 IU/day raise hypercalcemia risk. More is not better; sufficient is better.
Moderate evidence Omega-3 / fish oil
Modest cardiovascular support at ~1 g/day. High doses are prescription territory and carry a real atrial-fibrillation signal.
Omega-3 / fish oil
Modest cardiovascular support at ~1 g/day. High doses are prescription territory and carry a real atrial-fibrillation signal.
Modest cardiovascular support at ~1 g/day. High doses are prescription territory and carry a real atrial-fibrillation signal.
The omega-3 story is genuinely mixed. High-dose purified EPA cut cardiovascular events 25% in REDUCE-IT, while the similar STRENGTH trial with a different formulation found nothing — leaving even cardiologists arguing about mineral-oil placebos. At ordinary supplement doses (~1 g/day EPA+DHA), effects on triglycerides are real and effects on events are modest.
The honest downside: meta-analyses of the large trials show a dose-dependent increase in atrial fibrillation, mainly above 1 g/day. Claims about "biological aging" and telomeres rest on soft endpoints and should not drive a purchase.
How to take it: 1 g/day combined EPA+DHA — or simply eat oily fish twice a week. Buy fresh (check TOTOX/freshness where published); rancid fish oil is worse than none. A history of AF is a reason to talk to your doctor first.
Our pick
Life Extension Super Omega-3 — fish-oil concentrate tested by IFOS: two softgels give 1,200 mg of EPA and DHA, a little over the row’s 1 g; 120 softgels.
Buy at Life Extension Europe (opens lifeextensioneurope.com in a new tab)
Moderate evidence CoQ10
Real trial benefits — but in heart failure, as an add-on to treatment. General anti-aging and skin claims are unsupported.
CoQ10
Real trial benefits — but in heart failure, as an add-on to treatment. General anti-aging and skin claims are unsupported.
Real trial benefits — but in heart failure, as an add-on to treatment. General anti-aging and skin claims are unsupported.
CoQ10 has one context with genuinely impressive data: chronic heart failure. In the Q-SYMBIO randomized trial (420 patients, 2 years), 300 mg/day roughly halved major adverse cardiac events, and a meta-analysis of 14 trials found lower mortality (RR 0.69). That is "discuss with your cardiologist" evidence, not a reason for healthy people to supplement.
Statin muscle aches: evidence is genuinely mixed — some meta-analyses find modest relief, others none. A 100–200 mg/day trial for a few weeks is cheap and low-risk if statin aches bother you; keep expectations low.
Skin and "cellular energy" claims: no convincing outcome trials, including for the mitochondria-targeted analog MitoQ. Take with fat for absorption; mild caution with warfarin.
Emerging evidence Magnesium
Common intake shortfalls and a plausible sleep effect — but the sleep trials are few, small, and low-quality.
Magnesium
Common intake shortfalls and a plausible sleep effect — but the sleep trials are few, small, and low-quality.
Common intake shortfalls and a plausible sleep effect — but the sleep trials are few, small, and low-quality.
A meaningful share of European adults eats below the recommended magnesium intake, and status is hard to measure (serum misses tissue stores) — so supplementation is a reasonable, cheap hedge. The famous sleep benefit is thinner than its reputation: a meta-analysis in older adults found just three small trials showing sleep-onset ~17 minutes faster, with the evidence rated low-to-very-low quality. Newer positive trials of glycinate and threonate forms are sponsor-run.
How to take it: 200–400 mg elemental magnesium in the evening; glycinate or citrate are better tolerated than oxide. Threonate is expensive with the least outcome data.
Safety: loose stools at higher doses; caution with kidney impairment; separate from some antibiotics and bisphosphonates by a few hours.
- Best for
- Top pick: Pure Encapsulations Magnesium Glycinate
Limited evidence Multivitamins — memory maybe, lifespan no
One trial platform, testing one donated product, found slower cognitive aging in older adults. 390,000 followed for 20+ years found zero mortality benefit, and nothing shows a benefit for skin or aging.
Multivitamins — memory maybe, lifespan no
One trial platform, testing one donated product, found slower cognitive aging in older adults. 390,000 followed for 20+ years found zero mortality benefit, and nothing shows a benefit for skin or aging.
One trial platform, testing one donated product, found slower cognitive aging in older adults. 390,000 followed for 20+ years found zero mortality benefit, and nothing shows a benefit for skin or aging.
Both halves of this are worth knowing. Within the COSMOS trial platform, three cognition substudies plus a meta-analysis across >5,000 participants found a daily multivitamin slowed global cognitive aging by the equivalent of about two years over 2–3 years of use — a replicated, if modest, effect in adults 60+.
Against the longevity claim: an NIH analysis of three prospective cohorts — 390,124 healthy adults followed over 20 years — found no reduction in all-cause, cancer, or cardiovascular mortality from daily multivitamin use. It is inexpensive insurance for older adults with narrowing diets, not a longevity drug.
We grade it limited: the one positive finding rests on a single trial platform and a single product, the largest follow-up found no effect on how long people live, and no trial shows a multivitamin slowing skin or body aging in people who eat well.
Caveat: the cognition finding comes from one trial platform testing one product (Centrum Silver, donated by its manufacturer). Consistent within itself; not yet replicated elsewhere.
The longevity frontier
Moderate evidence Urolithin A (Mitopure)
Two well-run RCTs show better muscle endurance and mitophagy markers — with primary endpoints missed and all trials maker-funded.
Urolithin A (Mitopure)
Two well-run RCTs show better muscle endurance and mitophagy markers — with primary endpoints missed and all trials maker-funded.
Two well-run RCTs show better muscle endurance and mitophagy markers — with primary endpoints missed and all trials maker-funded.
Urolithin A is a gut metabolite of pomegranate compounds that only ~40% of people produce naturally — the plausible rationale for supplementing it directly. Two randomized trials exist, both decent, both funded by the manufacturer: in middle-aged adults, 4 months at 500–1,000 mg/day missed its primary endpoint (peak power) but produced ~12% leg-strength gains and lower inflammation markers; in older adults, muscle-endurance measures and mitophagy biomarkers improved while the primary endpoints did not.
The pattern — consistent secondary and biomarker wins, missed primaries — earns a cautious "moderate": likely doing something real to muscle mitochondria, magnitude modest, evidence base entirely industry-run.
How to take it: 500–1,000 mg/day; both trials ran 4 months. Safety was clean.
- Best for
- Top pick: Timeline Mitopure (the product used in both RCTs)
Emerging evidence NAD+ precursors (NMN & NR)
They reliably raise blood NAD+. No trial has yet shown an anti-aging outcome — and NMN is not legal to sell as food in the EU.
NAD+ precursors (NMN & NR)
They reliably raise blood NAD+. No trial has yet shown an anti-aging outcome — and NMN is not legal to sell as food in the EU.
They reliably raise blood NAD+. No trial has yet shown an anti-aging outcome — and NMN is not legal to sell as food in the EU.
NAD+ declines with age, and both NMN and NR raise blood NAD+ reliably in humans — that much is settled. What hasn't happened is the next step: human trials (250–900 mg/day NMN; 300 mg NR) show scattered, inconsistent wins on insulin sensitivity or walking distance, and a recent meta-analysis found no overall physical-function benefit in older adults. Biomarker up, outcomes pending.
Regulatory reality: in the EU, NMN is an unauthorized novel food — products sold here are grey-market. Nicotinamide riboside (Tru Niagen) went through EFSA assessment and is the legal option at 300 mg/day.
Safety: clean in trials to date. One flag worth knowing: NAD+ metabolism feeds rapidly dividing cells, and mouse data raise questions in active cancer — anyone with a cancer history should ask their oncologist first.
- Best for
- Top pick: Tru Niagen (nicotinamide riboside — the EU-legal precursor)
Emerging evidence Glycine + NAC (GlyNAC)
Striking results from one small lab; the one independent trial missed its primary endpoint. Watch, don't bet.
Glycine + NAC (GlyNAC)
Striking results from one small lab; the one independent trial missed its primary endpoint. Watch, don't bet.
Striking results from one small lab; the one independent trial missed its primary endpoint. Watch, don't bet.
GlyNAC supplies the precursors of glutathione, the cell's main antioxidant, which declines with age. A Baylor pilot and a 16-week RCT (24 older adults) reported an almost implausible breadth of improvements — oxidative stress, mitochondrial function, inflammation, insulin resistance, gait speed, strength, cognition. Tiny samples, one research group, everything-improved pattern: exactly the profile that demands replication.
The independent test so far: a 114-person trial failed its primary endpoint — no overall glutathione increase — with only a low-glutathione subgroup responding in post-hoc analysis (though its 14-day duration limits the conclusion).
If you experiment anyway: trials used ~100 mg/kg/day of each (roughly 7 g + 7 g). NAC is a drug in most of Europe — check interactions (notably nitroglycerin) with a pharmacist.
Emerging evidence Senolytics (dasatinib + quercetin, fisetin)
D+Q genuinely clears senescent cells in humans — but no senolytic has improved a clinical outcome, and shop-bought versions aren't the trial drugs.
Senolytics (dasatinib + quercetin, fisetin)
D+Q genuinely clears senescent cells in humans — but no senolytic has improved a clinical outcome, and shop-bought versions aren't the trial drugs.
D+Q genuinely clears senescent cells in humans — but no senolytic has improved a clinical outcome, and shop-bought versions aren't the trial drugs.
Senolytics aim to clear senescent "zombie" cells. Dasatinib + quercetin is the only regimen with demonstrated target engagement in humans — a phase-1 pilot reduced senescent-cell burden within days — and a 20-week phase-2 trial in postmenopausal women moved one bone-formation marker. That's it: no senolytic has improved a clinical aging outcome in a published RCT.
Fisetin is behind, not ahead: despite the hype, the Mayo Clinic frailty trials remain unpublished, and trial doses (~1.4 g for a 70 kg woman) dwarf label doses of shop supplements — which have zero published human efficacy data.
The safety framing matters: dasatinib is a chemotherapy drug with real off-target risks, dosed in short pulses under supervision. This category belongs in clinical trials, not shopping baskets. Note also that quercetin inhibits drug-metabolizing enzymes (CYP3A4) — it interacts with common medications.
Emerging evidence Low-dose rapamycin (prescription-only)
The strongest animal longevity drug ever tested. In humans: one year of safety data, a missed primary endpoint, exploratory wins.
Low-dose rapamycin (prescription-only)
The strongest animal longevity drug ever tested. In humans: one year of safety data, a missed primary endpoint, exploratory wins.
The strongest animal longevity drug ever tested. In humans: one year of safety data, a missed primary endpoint, exploratory wins.
Rapamycin extends lifespan more reliably than any other molecule in mouse programs — which is why physicians quietly prescribe weekly low doses off-label. The human evidence arrived in 2024–25 with PEARL, the largest longevity RCT to date (114 adults, 48 weeks): safe over the year, primary endpoint (visceral fat) unchanged, with women on the higher dose gaining lean mass and reporting less pain.
That is a defensible "emerging": genuine trial activity, plausible mechanism, no proven benefit yet. Known costs of the molecule: mouth ulcers, lipid elevations, glucose intolerance, immune effects.
Bottom line: a prescription medicine under investigation — reasonable to discuss with a physician, ideally inside a study; indefensible to self-source online.
Limited evidence Metformin for healthy non-diabetics
The flagship observational finding collapsed on re-analysis, the TAME trial has still not started, and it blunts training adaptations.
Metformin for healthy non-diabetics
The flagship observational finding collapsed on re-analysis, the TAME trial has still not started, and it blunts training adaptations.
The flagship observational finding collapsed on re-analysis, the TAME trial has still not started, and it blunts training adaptations.
Metformin's longevity case rested on a famous 2014 observation that diabetics on metformin outlived non-diabetics. A 2023 re-analysis found the result was an artifact of how patients left the metformin group — over 20 years, metformin-treated diabetics did worse than matched non-diabetics. The purpose-built TAME trial had still not started by mid-2026, its sponsor still raising the money to launch it (AFAR, archived).
Meanwhile, metformin measurably blunts some of the adaptations to exercise training in older adults — a strange trade for a healthy person whose best longevity tool is training.
Bottom line: for people without diabetes or prediabetes, there is no RCT evidence and a weakening observational case. If you have prediabetes, that's a different (and legitimate) conversation with your doctor.
Limited evidence Spermidine
The one substantial human trial was null. The supporting evidence is diet-cohort data confounded by healthy eating.
Spermidine
The one substantial human trial was null. The supporting evidence is diet-cohort data confounded by healthy eating.
The one substantial human trial was null. The supporting evidence is diet-cohort data confounded by healthy eating.
Spermidine induces autophagy in the lab and tracks with lower mortality in dietary cohorts — higher-spermidine diets meant a ~24% lower mortality hazard in the Bruneck study. But spermidine-rich diets are wholegrain-vegetable-aged-cheese diets: classic healthy-eater confounding.
The interventional test failed. SmartAge — 100 older adults, 12 months of wheat-germ spermidine at the marketed dose — found no effect on memory or biomarkers versus placebo, unwinding the promising 3-month pilot that preceded it.
Bottom line: eat the foods (whole grains, mushrooms, aged cheese, natto); skip the capsules until a positive human trial exists.
Limited evidence Resveratrol
The cautionary tale: spectacular mouse mechanism, terrible human bioavailability, two decades of trials without a clinical payoff.
Resveratrol
The cautionary tale: spectacular mouse mechanism, terrible human bioavailability, two decades of trials without a clinical payoff.
The cautionary tale: spectacular mouse mechanism, terrible human bioavailability, two decades of trials without a clinical payoff.
Resveratrol is why this guide insists on human outcomes. The red-wine molecule activated sirtuins in the lab, extended lifespan in some organisms (most reliably yeast and worms, per a 2012 meta-analysis), sold a company to GSK for $720 million — and then failed to deliver in human trials. It is destroyed by first-pass metabolism (oral bioavailability "considerably less than 1%", per a 2011 review), and a 2024 systematic review of almost 200 human studies found consistent falls in inflammatory markers but "no conclusive clinical evidence to advocate its recommendation in any healthcare setting"; an earlier review found "ambiguous and sometimes even detrimental effects" in some cancers and fatty liver disease. GSK shelved the program in 2010, after concluding that its resveratrol drug "may only offer minimal efficacy" while risking kidney complications in myeloma patients (FierceBiotech, 2010).
Bottom line: the marketed pills are pharmacologically implausible at their doses. If any sirtuin story survives, it will not be this molecule at this bioavailability.
Limited evidence Taurine
The 2023 "anti-aging" headline collapsed in 2025 — taurine doesn't even decline with age in most humans. Mild blood-pressure effects are real.
Taurine
The 2023 "anti-aging" headline collapsed in 2025 — taurine doesn't even decline with age in most humans. Mild blood-pressure effects are real.
The 2023 "anti-aging" headline collapsed in 2025 — taurine doesn't even decline with age in most humans. Mild blood-pressure effects are real.
In 2023, a Science paper reported taurine declines with age and that supplementing extended mouse lifespan ~10–12% — headlines followed. In 2025, a second Science paper using longitudinal data in humans, monkeys, and mice found circulating taurine rises or stays flat with age in most cohorts. The NIH's conclusion: taurine is unlikely to be a useful aging biomarker, and supplementing on that basis is premature.
What human trials do show is mild cardiometabolic support: meta-analyses report blood pressure ~4.7/2.9 mmHg lower and modestly better lipids at 1–3 g/day. Cheap and safe — just not anti-aging.
Beauty from within
Strong evidence Collagen peptides
The best-evidenced beauty supplement: pooled trials show gains in skin hydration, elasticity and wrinkles. Take about 10 g a day with vitamin C, alongside a retinoid.
Collagen peptides
The best-evidenced beauty supplement: pooled trials show gains in skin hydration, elasticity and wrinkles. Take about 10 g a day with vitamin C, alongside a retinoid.
The best-evidenced beauty supplement: pooled trials show gains in skin hydration, elasticity and wrinkles. Take about 10 g a day with vitamin C, alongside a retinoid.
Hydrolyzed collagen has the deepest trial base of any oral skin supplement — a 2021 meta-analysis of 19 trials and a 2023 meta-analysis of 26 show measurable improvements in skin elasticity, hydration, and wrinkle depth over 8–12 weeks, plus joint-stiffness (2019 osteoarthritis meta-analysis) and bone-density (12-month RCT) effects on longer horizons. The effects are modest, and most trials are funded by the ingredient makers. A 2025 meta-analysis of 23 RCTs found no effect in the few trials without pharmaceutical-company funding and concluded there is no clinical evidence, but that goes too far: two of those trials gave only 1–2 g a day, and the clearest negative trial measured forearm skin in women aged 60 to 93 with fragile, thinning skin (our reading of it).
Dose and partners matter: about 10 g a day, with vitamin C and other skin nutrients such as hyaluronic acid. The skin may also need a local signal to build collagen where you want it, such as a nightly retinoid or a collagen-stimulating treatment; collagen alone at a low dose may not be enough to see a result.
We maintain an entire evidence guide on it — mechanisms, forms, doses, safety, and how it stacks against clinical treatments: the collagen guide.
- Best for
- Top pick: Aeterna COLLAGEN MAX PRO (10 g Peptan® peptides per scoop)
Our pick
Aeterna COLLAGEN MAX PRO — 10 g of Peptan® collagen peptides per 13 g scoop, with a hyaluronic-acid matrix, plant ceramides and acerola vitamin C. Bovine, or fish in the Marin C+ version.
See the product Label facts, prices, the evidence graded.
Emerging evidence Astaxanthin
Small consistent trials show better skin moisture and elasticity and mild internal UV protection. Not a sunscreen replacement.
Astaxanthin
Small consistent trials show better skin moisture and elasticity and mild internal UV protection. Not a sunscreen replacement.
Small consistent trials show better skin moisture and elasticity and mild internal UV protection. Not a sunscreen replacement.
Astaxanthin is the carotenoid that makes salmon pink, and the best-studied of the "internal photoprotection" supplements. A 2021 meta-analysis (11 studies) found improved skin moisture and elasticity and reduced wrinkle measures; a 10-week RCT showed a higher UV dose was needed to redden supplemented skin.
The caveats are structural: trials are small (tens of participants), short (6–16 weeks), largely run in Japan by manufacturers, and measure instrument-level changes — visible-in-the-mirror is not the standard here.
How to take it: 4–6 mg/day with a meal containing fat. Benign; at very high carotenoid doses skin can tint slightly orange. It supplements sunscreen — it never substitutes for it.
Emerging evidence Oral hyaluronic acid
Better than its reputation: a 2025 meta-analysis of 7 RCTs found real hydration and wrinkle improvements. Small, industry-funded trials.
Oral hyaluronic acid
Better than its reputation: a 2025 meta-analysis of 7 RCTs found real hydration and wrinkle improvements. Small, industry-funded trials.
Better than its reputation: a 2025 meta-analysis of 7 RCTs found real hydration and wrinkle improvements. Small, industry-funded trials.
Oral HA was long dismissed ("it's just digested") — the 2025 literature says otherwise. A meta-analysis of 7 placebo-controlled RCTs (379 participants) found statistically significant improvements in skin hydration, elasticity, and wrinkle depth, and a 150-person double-blind RCT reported hydration and barrier improvements.
Strength of evidence is comparable to collagen's early days: small, mostly Asian, heavily manufacturer-funded trials with instrument endpoints. A reasonable add-on for dry, crepey skin at 120–240 mg/day — with expectations calibrated to "measurable", not "transformative".
The do-not-buy list
Limited evidence Proprietary "longevity blends"
We found no outcome trial of any proprietary longevity blend, and the label need not say how much of each ingredient you get.
Proprietary "longevity blends"
We found no outcome trial of any proprietary longevity blend, and the label need not say how much of each ingredient you get.
We found no outcome trial of any proprietary longevity blend, and the label need not say how much of each ingredient you get.
The commercial center of the longevity market is the multi-ingredient blend: a sprinkle of resveratrol, some NMN, "senolytic botanicals", a proprietary label hiding per-ingredient doses. Three problems compound: we found no proprietary longevity blend tested against a clinical outcome in PubMed (searched October 2026); the listed ingredients may sit below their (already unsuccessful) trial doses; and the proprietary label makes verification impossible by design — US rules require the blend's total weight but "not the amounts of the individual ingredients" (J Nutr, 2023). The nearest test is not encouraging: in a 6-month trial, ten popular supplements taken daily — resveratrol, quercetin, curcumin and fish oil among them — left arterial stiffness, blood pressure, lipids, glucose and inflammation markers unchanged.
If you want to experiment at the frontier, buy the single molecule at the trial dose from a third-party-tested brand — labels are not self-verifying: NMN and urolithin A products tested in 2024 ranged from 28.6% above to 100% below their label claims. A blend is a portfolio of underdosed bets sold at a premium.
Limited evidence "Detox" products
No defined toxin, no validated endpoint, no trials. Your liver and kidneys already do this job.
"Detox" products
No defined toxin, no validated endpoint, no trials. Your liver and kidneys already do this job.
No defined toxin, no validated endpoint, no trials. Your liver and kidneys already do this job.
Detox supplements name no specific toxin, measure no validated endpoint, and have no controlled trials: a 2015 critical review found "no randomised controlled trials" of commercial detox diets in humans, only a handful of small studies with flawed methods. The organs that clear metabolic waste (liver, kidneys) are not supplement-responsive filters, and "toxin buildup" as marketed does not correspond to a measurable clinical state. Some products carry real risk: a 2026 review of detox and diet teas found case reports of electrolyte disturbances, cardiovascular events and liver injury, and products with undeclared drugs.
The one legitimate nearby idea is reducing intake of harmful exposures: alcohol, smoking, ultra-processed food. That costs nothing.
Limited evidence Epithalon & injectable "peptide bioregulators"
Sold online for injection with essentially no independent human evidence and no quality control. Do not buy.
Epithalon & injectable "peptide bioregulators"
Sold online for injection with essentially no independent human evidence and no quality control. Do not buy.
Sold online for injection with essentially no independent human evidence and no quality control. Do not buy.
Epithalon (a synthetic tetrapeptide) and its "bioregulator" siblings are marketed online as telomerase-activating anti-aging injections. The human data come mainly from one St Petersburg research institute and its collaborators, and largely concern Epithalamin, the pineal extract epithalon was modelled on (2025 review) — such as a 2003 report of lower mortality in 266 older people — and we found no large, independent double-blind trial in PubMed (searched October 2026). A 2026 review describes unapproved peptides like it as having "limited clinical evidence" and lacking "long-term safety data". Products ship as "research chemicals" — a label that waives every quality standard. We found no published purity test of epithalon itself, but illicit peptide drugs have a documented record: a 2018 Belgian screen of falsified peptide drugs bought from suspected illegal online pharmacies found purity as low as 5% for some, plus arsenic and lead contamination. There is also a theoretical telomerase–cancer concern nobody has resolved: in a 2025 cell study, epithalon lengthened telomeres in two breast-cancer cell lines as well as in normal cells.
For an EU consumer this category is an unambiguous do-not-buy: unproven, unregulated, and injectable.
The starting routine
What you can do at home
Four steps, nothing exotic: the few supplements with the deepest human trials, and the food and training they work with. Add anything else only for a reason you can name.
-
Every day, with strength training
Creatine monohydrate, 3–5 g
Plain monohydrate, every day, no loading phase. In the trials its gains in muscle and strength came with resistance training, so pair it with two or three sessions a week. Expect about 1 kg of water in the muscle in the first weeks: hydration, not fat. The evidence.
Our pick
Thorne Creatine — creatine monohydrate and nothing else, 5 g a scoop, NSF Certified for Sport; 450 g.
-
At every meal
Protein and fibre, food first
25–40 g of protein a meal, about 1.2–1.6 g per kilo of body weight a day; fibre to 25–29 g a day from food, and 5–10 g of psyllium with plenty of water if you fall short. Protein · fibre.
-
Any time of day
Collagen peptides, about 10 g, with vitamin C
The dose of the skin trials, every day, with vitamin C and other skin nutrients such as hyaluronic acid. The skin may also need a local signal to build collagen where you want it, such as a nightly retinoid. The evidence.
Our pick
Aeterna COLLAGEN MAX PRO — 10 g of Peptan® collagen peptides per 13 g scoop, with a hyaluronic-acid matrix, plant ceramides and acerola vitamin C. Bovine, or fish in the Marin C+ version.
See the product Label facts, prices, the evidence graded.
-
Once, then again after 3–4 months
A vitamin D test before the vitamin D
A 25-OH-D blood test first; take 800–2,000 IU of D3 a day only if you are low, then retest after 3–4 months. Deficiency is common at European latitudes, especially from October to April; in people who were not deficient, supplements showed no benefit. The evidence.
Give collagen the 8–12 weeks of its skin trials before you judge it. Omega-3 is optional: oily fish twice a week, or 1 g a day and no more (why). With kidney disease, talk to your doctor before taking creatine.
Buying without being fooled
How to evaluate any supplement before buying
Look for batch-level third-party testing — NSF, Informed Sport, USP — and trial-matched doses on the label.
How to evaluate any supplement before buying
Look for batch-level third-party testing — NSF, Informed Sport, USP — and trial-matched doses on the label.
Supplements are regulated as foods, not medicines — nobody checks a product does what it claims before it reaches you. Three certification programs do independent verification:
- NSF Certified for Sport — every batch tested for ~280 banned substances, label accuracy, and contaminants; the only program recognized by USADA.
- Informed Sport / Informed Choice — UK-based (LGC) batch testing; the most relevant sports certification in Europe.
- USP Verified — verifies identity, strength, purity, and manufacturing quality (not banned substances).
In the EU, any ingredient without significant pre-1997 food history needs novel-food authorization (Regulation 2015/2283) — which is why nicotinamide riboside is legal here and NMN is not. "Anti-aging", "senolytic", and "NAD+ booster" are not EFSA-authorized health claims; a seller using them is already outside the rules.
Then check the dose: a product citing a trial should contain the trial's dose. A "longevity blend" with 50 mg of resveratrol cites studies that used 1,000+ mg — of a molecule that failed anyway.
Red flags that end the conversation
Proprietary blends, mouse citations, injectable "research chemicals", cure-all breadth, no certificate of analysis.
Red flags that end the conversation
Proprietary blends, mouse citations, injectable "research chemicals", cure-all breadth, no certificate of analysis.
Any one of these is a reason to close the tab:
- "Proprietary blend" without per-ingredient doses — structurally designed to hide underdosing.
- Mouse studies cited as if human — check whether the linked paper says "mice", "C. elegans", or "in vitro".
- Peptides or "research chemicals" sold for injection (epithalon, BPC-157, GH secretagogues) — unregulated, unproven, and with documented purity problems from grey-market vendors.
- Cure-all breadth — energy, skin, brain, immunity, and longevity from one capsule.
- No batch testing or certificate of analysis, or claims of being "FDA approved" (supplements never are).
- Subscription-trap pricing and countdown timers — evidence-confident brands don't need urgency theater.
Frequently asked questions
What single supplement has the best anti-aging evidence?
Creatine
What single supplement has the best anti-aging evidence?
Creatine
Boring but true: creatine (3–5 g/day) attached to resistance training, on a protein-sufficient diet, is the only "stack" with multiple meta-analyses behind it. Fiber at 25–29 g/day has the best mortality data of anything sold in a supplement aisle. Everything more exotic ranks below these on evidence.
Do NAD+ boosters actually reverse aging?
Not yet
Do NAD+ boosters actually reverse aging?
Not yet
NMN and NR reliably raise blood NAD+. No trial has shown that this translates into an anti-aging clinical outcome, and a 2025 meta-analysis found no physical-function benefit in older adults. If you want to try anyway, the EU-legal option is nicotinamide riboside at 300 mg/day. Treat it as a hopeful experiment, not a treatment.
Should I test my vitamin D or just take it?
Test first
Should I test my vitamin D or just take it?
Test first
Testing beats guessing: the large trials in people not selected for deficiency — including the NEJM fracture analysis of VITAL — were null, while correcting real deficiency remains standard care. A 25-OH-D test costs little; 800–2,000 IU/day is the safe correction range if you're low.
Are shop-bought "senolytics" the same as the ones in trials?
No
Are shop-bought "senolytics" the same as the ones in trials?
No
The trials pair dasatinib — a chemotherapy agent — with gram-level quercetin in short supervised cycles, and even those haven't shown clinical benefits yet. Shop-bought quercetin or fisetin at label doses has zero published human efficacy data. The name is shared; the intervention is not.
Is rapamycin or metformin worth asking my doctor about?
Discuss it
Is rapamycin or metformin worth asking my doctor about?
Discuss it
Both are prescription molecules under investigation, not supplements. Metformin's longevity case weakened badly after re-analysis of its flagship observational study, and its dedicated trial (TAME) had still not started by mid-2026. Rapamycin's PEARL trial was safe over 48 weeks but missed its primary endpoint. Reasonable to discuss with a physician — ideally inside a study. Unreasonable to buy online.
Do beauty supplements like collagen and oral HA really work?
Modestly
Do beauty supplements like collagen and oral HA really work?
Modestly
Better than their reputation, weaker than their marketing: collagen peptides, astaxanthin, and oral hyaluronic acid all have positive randomized trials and meta-analyses on hydration, elasticity, and wrinkle measures — small, short, and mostly industry-funded. Collagen has by far the deepest base (about 10 g a day, with vitamin C). Effects are instrument-measurable rather than transformative. Sunscreen and retinoids outrank them all; see the collagen guide for the full picture.
Do multivitamins do anything if I eat well?
Lifespan, no
Do multivitamins do anything if I eat well?
Lifespan, no
For mortality: 390,124 adults followed over 20 years showed no benefit. For cognition: the COSMOS platform found a modest, internally replicated slowing of cognitive aging in adults 60+. Cheap insurance for older adults with narrowing diets — not a longevity intervention.
Interactions & safety
Outside prenatal basics, the trials here excluded pregnant women, so in pregnancy and breastfeeding the answer is no without a doctor’s sign-off. On prescription medicines, ask a pharmacist first: fish oil, quercetin, NAC and magnesium all have interactions.
Interactions and safety flags
The recurring themes: fish-oil AF risk, vitamin D megadoses, NMN legality, quercetin drug interactions, kidney caution.
Interactions and safety flags
The recurring themes: fish-oil AF risk, vitamin D megadoses, NMN legality, quercetin drug interactions, kidney caution.
Individual sections carry their own cautions; these are the recurring ones worth a second read:
- Fish oil — dose-dependent atrial-fibrillation signal above ~1 g/day; bleeding risk at high doses with anticoagulants.
- Vitamin D — no monthly/annual megadoses; chronic intake above ~4,000 IU/day risks hypercalcemia.
- NAD+ precursors — NMN is not authorized as a food in the EU; anyone with an active or recent cancer should involve their oncologist.
- Quercetin (in "senolytic" products) — inhibits CYP3A4 and P-glycoprotein; interacts with many common medicines.
- NAC — interactions incl. nitroglycerin; it is a pharmacy medicine in much of Europe.
- Magnesium — caution in kidney impairment; separate from certain antibiotics.
- Metformin, rapamycin, dasatinib — prescription drugs with real adverse-effect profiles; never self-source.
- Pregnancy and breastfeeding — outside prenatal basics, the trials above excluded pregnant women; default to "no" without clinician sign-off.
Educational content, not medical advice. Talk to a qualified clinician before starting supplements, especially during pregnancy, breastfeeding, while on medications, or with a chronic condition.