Supplements
A short list of supplements has real human evidence, almost everything else is hype, and a handful cause harm at doses people commonly take. After hundreds of thousands of person-years of randomized data, no supplement has been shown to extend life or prevent disease in healthy middle-aged adults at anything like the scale of Mediterranean-pattern eating, regular exercise, decent sleep and not smoking.
Use supplements as gap-fillers, and test first for anything fat-soluble, mineral or taken for years. Five clear that bar for a generally healthy 35–55-year-old who exercises moderately: vitamin D3 at the standard dose through winter above about 40° latitude, more only if you are low, omega-3 (EPA and DHA, the two long-chain marine fats) if you don't eat fish, magnesium if your diet falls short, vitamin B12 if you are vegetarian, take metformin or are past about 60, and creatine if you lift weights. The firmest evidence is for correcting deficiencies; most claims beyond that rest on observational studies, surrogate markers or small, often industry-funded trials, and the sections below say which.
Test before you supplement
Supplementing blind is how people end up with high blood calcium from megadose vitamin D, copper depletion from chronic zinc, or nerve damage from "energy" formulas loaded with B6. One caveat: the Endocrine Society advises against routine 25-hydroxyvitamin D testing in healthy adults, and the US Preventive Services Task Force (USPSTF) finds the evidence insufficient to judge screening, because correcting an asymptomatic low reading hasn't changed hard outcomes in trials. Test to dose deliberately, not as a population screen.
These add to the general panel in Midlife labs:
- 25-hydroxyvitamin D — the most informative supplement-related lab.
- Vitamin B12 with methylmalonic acid (MMA) or holotranscobalamin (the B12 available to cells) — serum B12 alone is insensitive.
- Ferritin and transferrin saturation — especially premenopausal women (low iron) and men over 50 (occasionally high).
- Magnesium — serum levels are insensitive to body stores, and red-blood-cell magnesium is not an established better alternative; see Serum magnesium.
- Omega-3 index — EPA + DHA as a percentage of red blood cell membranes; the usual target is 8–12%.
- Homocysteine — a downstream readout of B12, folate and B6 status.
- Thyroid-stimulating hormone (TSH) in context — 4–6 mIU/L in an older adult is most likely physiological; see Thyroid management.
Vitamin D3: fix a deficiency, then stop
Vitamin D is well worth correcting if you are deficient, especially for bone; in people with decent levels, large trials found nothing. The largest, VITAL, gave 25,871 US adults 2000 IU/day for a median of 5.3 years with no effect on cancer, cardiovascular events or all-cause mortality.[1] The extras are weak leads: cancer deaths about 12% lower in the subgroup of trials that dosed daily, with the all-trials estimate null,[2] and new autoimmune disease 22% lower in VITAL,[3] a borderline result that shrank once late cases were counted and was gone two years after people stopped.[4]
- No guideline sets a target; roughly 50–125 nmol/L (20–50 ng/mL) keeps you out of deficiency. At 30–50 nmol/L, take 2000 IU/day for 12 weeks; below 30, 4000 IU/day for 8–12 weeks; then retest. In range and under 75, nothing above the standard 600–800 IU/day has been shown to help.
- Daily, not large weekly, monthly or yearly doses, which are at best useless and at worst harmful. A meal helps absorption but is not essential.
- 4000 IU/day is the official upper limit, a cautious ceiling rather than a toxic dose; above it for months, check blood calcium.
- Beyond about 40° latitude, winter sun makes no vitamin D for part of the year, and stores last only weeks, so a good summer cannot cover the winter.
Omega-3: eat fish, or take a modest dose
Omega-3 reliably lowers triglycerides; for heart outcomes the evidence is weak to moderate and best in people who eat little fish. The one clear high-dose success, REDUCE-IT, gave 8,179 statin-treated adults with high triglycerides 4 g/day of prescription purified EPA (icosapent ethyl) and cut major cardiovascular events by 25%, though how much its mineral-oil placebo flattered that is disputed.[5] STRENGTH, at the same dose of mixed EPA+DHA, found no benefit,[6] and neither did the low-dose omega-3 arm of VITAL, whose benefit in low-fish eaters was a secondary finding.[7]
DO-HEALTH gave the first randomized signal that 1 g/day slows several DNA-methylation ageing clocks, but in a post-hoc analysis of 777 of its 2,157 participants, all over 70, in a trial that missed all six pre-specified primaries; clocks are surrogates, and vitamin D on its own moved none of them.[8]
- Two or more servings of fatty fish a week makes supplementation unnecessary.
- Without fish, 1–1.5 g/day of combined EPA+DHA is a reasonable personal default that lifts most people's omega-3 index toward 8%. No guideline recommends it: the 2026 US cholesterol guideline rates over-the-counter supplements "no benefit" for blood lipids and reserves omega-3 for prescription icosapent ethyl in high-risk patients.[9] Algae oil supplies the same fats for vegans; choose the form on price, since both common forms absorb well with a meal.
- Atrial fibrillation rises with dose. Odds were about 40% higher above 1.5 g/day in people at high cardiovascular risk, with no clear excess at 1.5 g/day or less,[10] though an earlier pooling saw a small 12% rise even at 1 g/day or less.[11] Going higher is a deliberate trade, and 4 g/day is prescription territory.
- Refrigerate after opening and look for an International Fish Oil Standards (IFOS) or Labdoor mark. Half of 171 North American products exceeded at least one voluntary oxidation limit,[12] and in a small trial oxidised oil lost the cholesterol benefit fresh oil delivered.[13]
Magnesium: a repletion mineral, not a drug
Topping up a low intake of magnesium is well supported; just under half of Americans take in less than the median requirement, a measure of intake rather than deficiency.[14] Its best-tested benefit is modest: pooling 34 trials at a median of about 368 mg/day, blood pressure fell about 2/2 mmHg.[15] A 2025 update of 38 trials found the effect concentrated in people already treated for hypertension (about 7.7 mmHg systolic, from just six trials) and in the deficient, with no clear effect in people with normal blood pressure.[16]
Drinking water counts. Over 16 years in 26,733 postmenopausal Swedish women, tap water richer in calcium and magnesium (magnesium about 10 versus 5 mg/L) went with about 13% lower risk of ischemic stroke, and the top third of water magnesium against the bottom third with about 31% lower; only the magnesium carried the link.[17] Unremineralised reverse-osmosis water loses it; see Water.
| Form | Best use | Notes |
|---|---|---|
| Glycinate | Evening, daily | Better tolerated than oxide in one small trial; no absorption edge shown |
| Citrate | Daily, mild constipation | Well absorbed; mildly laxative |
| Threonate | Cognition (preliminary) | Brain uptake shown only in animals; expensive, little elemental magnesium |
| Malate / taurate | General use | Fibromyalgia (malate) null in its one blinded trial; taurate's blood-pressure use is plausibility only |
| Oxide | Constipation only | Poorly absorbed |
- 200–400 mg/day elemental, but the top of that range is above both ceilings for supplemental magnesium: 250 mg/day in Europe and 350 mg/day in the US. Food magnesium has no limit.
- Be cautious with chronic kidney disease; separate doses by 2 hours from quinolone and tetracycline antibiotics and bisphosphonates.
Vitamin B12: essential for vegans, worth checking for a few others
Repletion in at-risk groups is well established; for healthy people who eat animal foods, routine B12 adds little. Vegetarians and vegans should take it without waiting for a test. Metformin users absorb it less well — each extra year on the drug was linked to about 13% higher odds of low B12 — and should consider periodic checks.[18] Long-term acid-suppressant users and adults past about 60 should get checked if tiredness or tingling appear; atrophic gastritis, which reduces the stomach acid and intrinsic factor B12 needs, affects about 8–9% of over-65s. In US survey data about 4% of adults had clearly low B12 and about 1 in 8 were borderline.[19] A cheap low-dose supplement is an alternative to testing.
- 25–250 µg/day for prevention; 1000 µg/day for documented deficiency. Cyanocobalamin works as well as methylcobalamin.
- Keep vitamin B6 under 12 mg/day, food and pills combined, the European upper limit. Chronic intake above it can cause peripheral neuropathy (tingling and numbness in the hands and feet), usually reversible; check "energy" B-complex labels.
The brain case is narrow: VITACOG enrolled adults with mild cognitive impairment and high homocysteine, measured brain shrinkage rather than thinking, and its "needs omega-3" finding is an unreplicated subgroup. The heart case is mostly closed: trials that lowered homocysteine did not reduce heart attacks or cardiovascular death, and cut stroke only slightly, mainly where food is not fortified with folic acid. More in B vitamins.
Creatine: muscle, if you lift
Creatine with resistance training adds muscle and strength, which is well established, though the extra gain past 60 is smaller and less certain. For cognition the one real lead is acute sleep loss, from two small trials by one group; in well-rested adults the evidence is weak. Large trials in Parkinson's, Huntington's and amyotrophic lateral sclerosis were null or stopped for futility.
Its safety record is unusually deep: across 685 trials and nearly 13,000 participants, some taking it for up to 14 years, side effects occurred in 4.6% on creatine and 4.2% on placebo.[20] About 70% of adults over 65 get under 1 g/day from food, which is why some researchers now propose calling it a conditionally essential nutrient; no official intake requirement exists.[21]
- 3–5 g/day of monohydrate, any time, no loading. Other forms cost 2–5× as much with no advantage; gummies often hold little creatine. Look for NSF Certified for Sport or USP Verified.
- Mention it before blood tests: creatinine runs slightly high and estimated kidney filtration can read low, without kidney damage. With chronic kidney disease, ask your nephrologist first.
- No "brain dose" has been established, and it doesn't help bone: a two-year trial in 237 postmenopausal women found no effect on bone density.[22]
Timing matters less than consistency for all five, since they work by filling stores: D3, omega-3 and any B12 with breakfast, magnesium glycinate 1–2 hours before bed, creatine whenever, with food.
Sleep and anxiety supplements: small nudges at best
The evidence is weak even for the best of them, and none replaces cognitive behavioural therapy for insomnia (CBT-I). Magnesium glycinate (200–400 mg) and L-theanine (200 mg: small self-rated sleep gains, nothing for anxiety) have the best risk-benefit. A 2026 review of 12 trials concluded magnesium is not supported as a routine insomnia treatment,[23] though in the best recent trial, of 250 mg/day as bisglycinate in 155 poor sleepers, about one extra person in seven improved meaningfully.[24]
Melatonin has a narrow place, dosed to the purpose: 0.5–5 mg of the immediate-release form at the destination's bedtime for jet lag, a small 0.3–1 mg dose to shift delayed sleep timing, and older adults with insomnia. For general insomnia the American Academy of Sleep Medicine advises, conditionally, against it.[25] A confounded 2025 observational report linked a year or more of use to higher heart-failure and death rates, a reason for strategic rather than indefinite nightly use.[26]
Ashwagandha, glycine, valerian and cannabidiol (CBD) are weak to preliminary; ashwagandha carries rare liver-injury reports, and CBD interferes with the liver enzymes that clear most prescription drugs. None causes meaningful addiction, tolerance or withdrawal at studied doses, and unlike over-the-counter antihistamines they don't stop working within days. See Sleep and anxiety supplements and the full ranking in Sleep supplements.
Longevity molecules: mouse data, thin human proof
In people the whole geroprotector category is weak or preliminary, and the gap between mechanism and clinical results has widened over five years, with industry funding distorting the literature.
Rapamycin extends mouse lifespan 10–25% across replicated trials. Related drugs improved older adults' immune response in small early trials,[27] but a 1,000-person phase 3 trial found no fewer respiratory illnesses,[28] and PEARL, the largest and longest trial to date of intermittent low-dose rapamycin in healthy adults, missed its primary endpoint over 48 weeks.[29] No human trial has hard longevity endpoints.
Urolithin A has the best nutraceutical data, mostly industry-funded: in 88 overweight middle-aged adults, hamstring strength rose about 10–12% against a 10% fall on placebo, but a missed primary endpoint of peak power.[30] Spermidine had 39% lower all-cause mortality in the top third of dietary intake in one town's long-running cohort,[31] but its largest supplement trial, 12 months in 100 older adults, missed its cognitive endpoint.[32]
NMN and nicotinamide riboside raise blood NAD⁺ (the cofactor cells use for energy metabolism) with small, inconsistent payoffs: better muscle insulin sensitivity in postmenopausal women with prediabetes,[33] lower systolic pressure only in a post-hoc subgroup,[34] no better than walking alone in high blood pressure,[35] and nothing at 2000 mg in obese men.[36] They may also raise NAD⁺ breakdown products that track with cardiovascular events in cohort data.[37] Resveratrol has effectively no human evidence at supplement doses, and lithium orotate no randomized trial in healthy adults.
The rest of the shelf
Curcumin: modest for joints, a rare liver risk. For mild-to-moderate knee osteoarthritis it beats placebo and matched a non-steroidal anti-inflammatory drug in the few head-to-head trials, on mostly low-certainty evidence;[38] after hard exercise it blunts soreness without improving performance.[39] The "reverses epigenetic age" claim comes from a whole-diet trial. Supplements, mostly high-absorption capsules, cause unpredictable liver injury that doesn't track dose, starting 1–4 months in and strongly linked to the gene variant HLA-B*35:01; probably very rare, roughly 1 in 10,000 to 1 in 100,000 exposed.[40] Not a daily anti-aging pill; a short course is no protection; stop at the first liver symptom.
Glucosamine: an association, not a cause. UK Biobank users had about 15% lower all-cause and 18% lower cardiovascular mortality,[41] a companion analysis about 22% lower cardiovascular death,[42] but that is about what healthy-user bias produces alone; it vanished in a larger, better-adjusted US analysis,[43] and no trial has tested it. For knee osteoarthritis, the publicly funded GAIT trial and an independent network meta-analysis were negative; the positive trials nearly all used one prescription product.[44][45] It can raise INR (clotting time) on warfarin.
Calcium: food first, never as a bolus. The highest total calcium intakes went with the lowest risk of new arterial calcification over 10 years.[46] A single pill of about 1,000 mg spikes blood calcium for hours, and supplements track with a probable rise in heart attacks; the evidence is contested, but pills offer no heart upside and a plausible downside food lacks. They don't measurably prevent fractures in healthy adults: across more than 51,000 community-dwelling adults over 50, calcium, vitamin D or both had no effect in any subgroup,[47] and the benefit is uncertain even in the deficient and care-home elderly. Aim for 1,000–1,200 mg/day from food (spinach doesn't count), supplement only the gap, no more than about 500 mg at once, with a meal; citrate suits acid-reducer users.
Vitamin K2: good mechanism, unproven outcomes. The menaquinone-7 (MK-7) form activates proteins that steer calcium into bone and out of arteries, but trials measure scans and markers, and they are split. For: 180 µg/day for three years improved arterial stiffness in postmenopausal women,[48] a 2026 trial slowed coronary calcium progression,[49] and another in 398 people with heavy coronary calcification slowed build-up by about a fifth.[50] Against: no effect on heart-valve calcification despite moving the biomarker,[51] and in kidney disease the biomarker moves while scans don't. It directly opposes warfarin; direct oral anticoagulants are not expected to be affected.
CoQ10: narrow and real, oversold elsewhere. It is genuinely disease-modifying in primary CoQ10 deficiency, a rare inherited disorder. In heart failure it probably helps, but the meta-analysis putting deaths about a third lower (relative risk 0.64) pooled mostly small, low-quality trials, and the US guideline does not recommend it.[52] KiSel-10, in Swedes aged 70–88 given CoQ10 with selenium, roughly halved cardiovascular deaths, but in a cohort that turned out to be low in selenium; it and the main heart-failure trial both had industry sponsorship. In the general population, users showed no mortality benefit, and those with obesity about 45% higher mortality in an exploratory observational subgroup.[53] Migraine prevention has a modest signal from small trials. Use plain ubiquinone with a fatty meal.
Supplements that do harm at common doses
Each was recommended on observational data or mechanism until randomized trials found harm. Pooling 78 trials in almost 300,000 people, antioxidant supplements did not prevent death, and in the low-bias trials beta-carotene, vitamin E and higher-dose vitamin A went with slightly higher mortality.[54]
- Vitamin E at 400 IU/day or more: about 39 extra deaths per 10,000 people,[55] more heart-failure hospitalisations in people with vascular disease or diabetes,[56] and 17% more prostate cancer in men.[57] Keep supplemental vitamin E to about 30 IU.
- Beta-carotene if you smoke or ever have: 18% more lung cancer in male smokers,[58] and a second trial stopped early for a 28% increase.[59] The USPSTF recommends against vitamin E or beta-carotene to prevent heart disease or cancer (Grade D).[60]
- Preformed vitamin A above 3000 µg/day (retinol activity equivalents): retinol intakes of 2,000 µg/day or more went with about 90% more hip fractures in nurses,[61] and high doses are teratogenic in pregnancy.
- High-dose calcium pills (above about 1,000 mg/day) without a deficiency: 31% more heart attacks with calcium taken without vitamin D in one meta-analysis,[62] and 17% more kidney stones with calcium plus vitamin D in the Women's Health Initiative.[63]
- High-dose antioxidants around training: in one short trial in young men, 1,000 mg vitamin C plus 400 IU vitamin E daily abolished the training gain in insulin sensitivity, a result a later 12-week trial did not reproduce.[64] Avoid vitamin C above 2,000 mg/day and selenium at 200 µg/day or more if you already get enough.
- Iron without documented deficiency, which is linked to diabetes and heart disease; men and post-menopausal women shouldn't supplement unless ferritin is low.
- Zinc above 40 mg/day long term, which depletes copper, and red yeast rice, an unlabelled statin.
Supplements to avoid has the thresholds and full citations.
Buying: quality is the weak link
Independent testing routinely finds 10–30% of products underdosed, contaminated with heavy metals or mislabelled.[65] Look for a USP Verified, NSF Certified for Sport or Informed Sport mark, and for fish oil, a tested oxidation (TOTOX) value. Melatonin is the worst offender: across 31 US products, content ranged from 83% below to 478% above the label, varying up to 465% between batches of one product;[66] prolonged-release prescription melatonin (Circadin 2 mg in the EU) is far more reliable. NMN products have repeatedly failed label-claim assays.
Optional add-ons
Situational, with a narrower case than the core five.
| Supplement | Dose | When it makes sense |
|---|---|---|
| Multivitamin (no iron) | Daily, optional | A gap-filler for a poor diet, on weak evidence: a modest cognitive-test gain in one trial (COSMOS-Mind), against a USPSTF "insufficient evidence" rating[67] |
| Folic acid | 400 µg/day | Women who could become pregnant |
| Iron (ferrous sulfate, alternate days) | As needed | Premenopausal women with ferritin <30 µg/L; bisglycinate if sulfate upsets your stomach (better tolerated, but not the form tested) |
| Vitamin K2 (MK-7) | 90–180 µg/day | Alongside D3 or a calcium supplement |
| CoQ10 | 300 mg/day in heart-failure and migraine trials; 100–200 mg/day otherwise | With your doctor, for a narrow reason above; statin muscle aches are contested |
| Curcumin | As studied for the formulation | Knee pain or training soreness; stop at the first liver symptom |
There is no case for taking everything at once. Prescription drugs, including glucagon-like peptide-1 (GLP-1) receptor agonists and testosterone therapy, live under Clinical Care.
For the full ordered action list drawn from every pillar, see the healthspan long list.