Four earn a place for most people: creatine, vitamin D if you are deficient, omega-3 if you eat little oily fish, and protein powder only as a convenience. A further group is worth taking under specific conditions. Most of the rest sit on rodent data or surrogate endpoints. Two, vitamin E and beta-carotene, have trial evidence of harm and should be actively avoided.
How these supplements were sorted
One question decides each verdict: is there human outcome evidence, or is there something less than that being presented as though it were?
A supplement earns the top bucket when randomised human trials show it does what it claims. It drops when the evidence is a surrogate marker, an association, or a mouse. It drops to the bottom when trials found harm.
This page ranks. It does not re-argue the individual cases, because each has its own entry and those entries carry the detail.
The supplement ranking
| Verdict | Supplement | In one line |
|---|---|---|
| Worth it | Creatine | The best-evidenced supplement in the list, and cheap |
| Worth it | Vitamin D3 | If you are deficient; correcting a deficiency is the benefit |
| Worth it | Omega-3 | If you eat little oily fish, and lowers triglycerides at dose |
| Worth it | Whey protein | Convenience, not magic; food does the same job |
| Conditional | Magnesium | Common shortfall; form matters more than dose |
| Conditional | Vitamin B12 | Vegans, over-65s, metformin and PPI users |
| Conditional | Zinc | Narrow window; excess causes its own problems |
| Conditional | Melatonin | Circadian timing, at far lower doses than sold |
| Conditional | Curcumin | Absorption is the whole problem |
| Not yet | NMN | Raises NAD+, a surrogate; no outcome data |
| Not yet | Spermidine | Association without trial confirmation |
| Not yet | Fisetin | Mouse results, human doses unestablished |
| Not yet | Taurine | Striking animal data, species gap unresolved |
| Not yet | Resveratrol | Human doses nowhere near the extrapolated ones |
| Not yet | Berberine | Real glucose effect, marketed well beyond it |
| Not yet | CoQ10 | Defensible on statins, thin otherwise |
| Avoid | Vitamin E and beta-carotene | Trials found increased harm, not benefit |
Verdicts are assigned by whether human outcome evidence exists. Supplements drop when the evidence is a surrogate marker, an association or an animal study, and drop to avoid where trials found harm.
What earns a place in the top group
Creatine is the outlier in this list: hundreds of human trials, a consistent effect on strength and lean mass, an emerging cognitive literature, and a cost of pennies a day. If you take one thing, it is this one.
Vitamin D earns its place conditionally on a blood test. The VITAL trial found no cardiovascular or cancer benefit in a largely replete population, which is the correct result to take from it: the benefit is in correcting deficiency, not in topping up someone who is fine.
Omega-3 makes sense if oily fish is rare in your diet, and separately has a genuine dose-dependent effect on triglycerides.
Whey is on the list for honesty rather than enthusiasm. It is a convenient way to reach a protein target, and nothing about the powder beats the same protein from food.
Conditional on your situation
These five are worth taking if a specific thing is true of you, and pointless otherwise. Magnesium if your intake is genuinely low. B12 if you are vegan, over 65, or on metformin or a proton pump inhibitor. Zinc only to correct a shortfall, since the useful window is narrow. Melatonin for circadian timing rather than sedation. Curcumin only in a formulation that addresses absorption.
The pattern is that each answers a question about you, not a question about longevity.
Promising, not proven
Seven supplements sit here, and they are the ones the longevity market is loudest about. They share a structure: a real mechanism, genuine scientific interest, and a gap between what has been demonstrated and what is being sold.
NMN raises NAD+, which is a surrogate rather than an outcome. Fisetin and taurine rest on animal work with unresolved species gaps. Spermidine has association without trial confirmation. Resveratrol's famous results came from doses no human takes. Berberine does affect glucose and is marketed far past that. CoQ10 is defensible alongside a statin and thin otherwise.
None of this makes them useless. It makes them experiments you are paying to run on yourself, which is a legitimate choice made knowingly and a poor one made because a label implied more.
The two to actively avoid
Vitamin E and beta-carotene are the only entries here where the evidence points at harm rather than absence of benefit, and the reason is specific.
The ATBC trial gave beta-carotene to male smokers and found lung cancer incidence went up. CARET tested beta-carotene with vitamin A and was stopped early, having found the same direction of effect. These were large randomised trials designed to demonstrate prevention, and they demonstrated the opposite.
The lesson generalises beyond these two compounds: an antioxidant that looks protective in observational data can be harmful when isolated, concentrated and taken daily. Dose and context are not details.
The four that earn a place
For most people the defensible stack is creatine, vitamin D if a test says so, omega-3 if fish is rare, and protein powder if it helps you hit a target. That is a short list against everything tracked here, and the shortness is the finding rather than an oversight.
Money not spent on the rest buys an ApoB test several times over, which will tell you something you do not currently know.
Frequently asked
Should I take all four of the top group?
Only the ones whose condition applies to you. Creatine is the one with the broadest case. Vitamin D depends on being deficient, omega-3 on eating little oily fish, and protein powder is a convenience rather than a requirement.
What if the supplement I take is not on this list?
The list covers what is tracked here, which is the compounds with enough human evidence to have a verdict either way. If something is missing, that usually means the evidence base is too thin to rank rather than that it has been dismissed.
Is “conditional” a polite way of saying no?
No. Conditional means the evidence is real but applies to a specific situation. Vitamin D corrects a deficiency and does very little otherwise. That is a genuine recommendation with a condition attached, not a hedge.
How often does this ranking change?
Rarely at the top and occasionally in the middle. The compounds with decades of trial evidence behind them do not move. The promising-but-unproven group is where a single good trial would change a verdict.
What about a multivitamin?
It is not on the list because it is not one compound. A multivitamin is a bundle, and the case for each ingredient is the case made in its own entry. Buying the bundle mostly buys the ingredients you did not need.
Where should I start if I take nothing?
With the question of whether you are deficient in anything, which needs a blood test rather than a purchase. After that, creatine is the one with the fewest conditions attached.
Sources
Each verdict rests on the evidence set out in that supplement’s own entry, which is linked from the table. The references below cover the claims made on this page specifically.
- The Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study Group. The effect of vitamin E and beta carotene on the incidence of lung cancer and other cancers in male smokers. New England Journal of Medicine, 1994. Beta-carotene supplementation increased lung cancer incidence.
- Omenn GS, Goodman GE, Thornquist MD, et al. Effects of a combination of beta carotene and vitamin A on lung cancer and cardiovascular disease. New England Journal of Medicine, 1996. The CARET trial was stopped early for harm.
- Kreider RB, Kalman DS, Antonio J, et al. International Society of Sports Nutrition position stand: safety and efficacy of creatine supplementation. JISSN, 2017.
- Manson JE, Cook NR, Lee IM, et al. Vitamin D supplements and prevention of cancer and cardiovascular disease (VITAL). New England Journal of Medicine, 2019. No reduction in either primary endpoint in a largely replete population.