Do NMN and NAD+ boosters do anything? Sorting the trials by how hard the endpoint was

Most people think NMN is the longevity supplement that finally has the science behind it. It does have the science — that’s the awkward bit. There are now dozens of human trials, several meta-analyses, and a genuinely impressive pile of biochemistry. And when you line those trials up in the right order, they tell a story nobody selling NMN wants told.

Here’s the order that matters: sort every trial by how hard the primary endpoint was. Not by result, not by date, not by dose. Just by how difficult it would be to fake, flatter or wobble the measurement. Do that, and the effect shrinks smoothly to nothing as the endpoints get harder.

That’s the whole post, really. But it’s worth walking through properly, because the individual studies are more interesting than the summary.

The claim, as everyone states it

The pitch is unusually coherent for a supplement, which is part of why it sells. It goes:

NAD+ is a coenzyme your cells need for basically everything — energy production, DNA repair, the sirtuin enzymes that got labelled “longevity genes” about fifteen years ago. NAD+ levels fall as you age. Restore them and you restore the cellular machinery that’s been quietly running out of fuel. NMN and NR are the precursors that let you do that with a capsule.

Every step of that is true except the last one, and the last one is the entire product.

Note the family resemblance to the collagen argument — a real decline, a real molecule, and a “therefore” doing an enormous amount of unexamined work in the middle. It’s the standard shape of a supplement story: the mechanism is genuinely correct right up until the point where it has to produce an outcome in a human being.

Where it came from

Not from nowhere, and not from marketing. This one came out of proper labs.

The modern NAD+ story descends from the sirtuin research of the 2000s, which descended from calorie restriction — the idea being that CR works partly by boosting NAD+ and activating sirtuins, so you might get the benefit without the hunger. (The one human CR trial we have, CALERIE, is worth reading precisely because it shows how modest even the real thing is.)

Then came two 2016 papers that lit the fuse. Zhang and colleagues in Science gave aged mice nicotinamide riboside, rejuvenated muscle stem cells, and — the headline — increased mouse lifespan. Mills and colleagues in Cell Metabolism ran twelve months of NMN in normally-aging mice and found suppressed age-related weight gain, better insulin sensitivity, better energy metabolism, more physical activity, even improved eye function, with no obvious toxicity. (According to PubMed: Zhang et al., DOI; Mills et al., DOI.)

Those are good papers. If you’d read them in 2016 you’d have been right to be excited. A supplement industry appeared almost immediately, and it has been running on those two mouse studies ever since — which is the tell, because ten years of human trials have happened in the meantime and the marketing still leads with the mice.

The endpoint-hardness ladder

Right, so here’s the framework. Four rungs, each one harder to fudge than the last.

Rung one: blood NAD+. A biochemistry readout. This works, unambiguously. Martens and colleagues in Nature Communications (2018) ran a 2×6-week crossover in healthy middle-aged and older adults on 1000 mg/day NR: well tolerated, NAD+ metabolism reliably stimulated. Yi and colleagues in GeroScience (2022) gave 80 healthy 40-65 year-olds placebo, 300, 600 or 900 mg of NMN for 60 days and got a clean dose-response in blood NAD+, highest in the 600 and 900 mg arms. Remie and colleagues in the American Journal of Clinical Nutrition (2020) confirmed it in muscle tissue itself, not just blood. (According to PubMed: Martens et al., DOI; Yi et al., DOI; Remie et al., DOI.)

So: the pill does the thing on the label. Genuinely. Hold onto that, because it’s the reason the whole category has such an honest-feeling story.

Rung two: soft functional outcomes. Six-minute walk distance. Questionnaires. Ventilatory thresholds during training. Here the results are positive — and here is where the industry money is.

That same Yi trial found significantly greater walking-distance gains in all three NMN arms versus placebo at 30 and 60 days, plus better SF-36 self-reported health scores. Liao and colleagues in the Journal of the International Society of Sports Nutrition (2021) put 48 amateur runners through six weeks of training on 300, 600 or 1200 mg NMN and found higher ventilatory thresholds in the middle and high-dose groups. (According to PubMed: Liao et al., DOI.)

Both of those are real trials with real placebo groups. But look at what they measure. A six-minute walk test is a how hard did you try test as much as a fitness test. SF-36 is you, ticking boxes about how you feel, having been given a capsule you may well believe in. And the Yi trial’s author list runs through an NMN manufacturer, an NMN chemicals company and a contract research organisation.

Worth flagging one more thing on that trial, because it’s the bit that quietly does the most marketing work: it also reported that “blood biological age” rose in placebo and stayed flat on NMN. That biological age came from the Aging.Ai 3.0 calculator — a free online tool that regresses a routine blood panel against age. It is not a methylation clock, and it is nowhere near the top of any honest ranking of biological age tests. “NMN stopped my biological age rising” sounds like a molecular claim. It’s a blood-panel calculator drifting a bit.

Rung three: objective physiology. Clamps, biopsies, spectroscopy, dynamometers — things that don’t care what you believe. This rung is where the effect dies.

Three separate reports from the same Danish 12-week trial (40 obese, insulin-resistant men on 2000 mg/day NR) found: no improvement in insulin sensitivity by hyperinsulinaemic-euglycaemic clamp, no change in endogenous glucose production, lipolysis, resting energy expenditure or body composition (AJCN, 2018); no change in skeletal muscle mitochondrial respiration, content or morphology (Journal of Physiology, 2020); and no change in β-cell function, glucagon or incretin hormones (JCEM, 2019). Remie’s Dutch crossover, with MR spectroscopy and muscle biopsies, raised muscle NAD+ metabolites and then found nothing for insulin sensitivity, mitochondrial function, liver fat, cardiac energy status or ambulatory blood pressure. Akasaka and colleagues in Geriatrics & Gerontology International (2023) gave older diabetic men with measurably impaired function 250 mg NMN for 24 weeks and found no difference in grip strength or walking speed — and that trial had NMN-industry authors on it too, which makes the null more striking, not less. (According to PubMed: Dollerup et al. 2018, DOI; Dollerup et al. 2020, DOI; Dollerup et al. 2019, DOI; Akasaka et al., DOI.)

One genuine exception, and it deserves its due. Yoshino and colleagues in Science (2021) gave 25 overweight, prediabetic postmenopausal women 250 mg NMN for ten weeks and found increased insulin-stimulated glucose disposal by clamp, with matching changes in muscle insulin signalling. That is a hard endpoint, an independent academic group, and a positive result. It is also n=25, one narrow population, ten weeks, and a measure of insulin signalling rather than anything about aging. (According to PubMed: Yoshino et al., DOI.) It’s the best card in the deck and it’s a fairly small card.

Rung four: aging outcomes. Mortality, disease incidence, disability, how long anyone lived. Nobody has measured this. Not once, in any trial, for any NAD+ precursor, in a human being. The entire premise of the category is untested.

What the meta-analyses do to the ladder

Predictably, they split along the same seam.

Wang and colleagues in Current Pharmaceutical Biotechnology (2025) pooled nine trials and 412 middle-aged and elderly participants and reported a significant effect on gait speed and a small drop in ALT, concluding NMN is “encouraging.” (According to PubMed: Wang et al., DOI.) Fair enough — that’s the strongest pooled case for the defence. Though note that 412 people across nine heterogeneous studies is a thin pool, and the paper reports its standardised mean difference for gait speed as “0.34 m/s,” which is a units error — an SMD is dimensionless. Small thing. Not confidence-inspiring.

Against it, Yang and colleagues in Nutrients (2026) ran a larger review: fifteen randomised trials, doses from 250 to 2000 mg/day, durations from 14 days to 24 weeks, GRADE assessment throughout. Findings: NMN is well tolerated, doesn’t raise adverse events, doesn’t disturb liver enzymes — and shows no significant effect on body weight, BMI, fasting glucose, HbA1c, lipid profile or systolic blood pressure. Diastolic pressure dropped slightly; HOMA-IR trended down without reaching significance. Their own summary is that “broad metabolic benefits were not evident.” (According to PubMed: Yang et al., DOI.)

Ten years, fifteen trials, and the honest headline is safe, and no clear metabolic benefit. On the ladder, that’s rungs one and two holding and rung three still empty.

The honest verdict: it does what it says, and what it says isn’t the point

So where does that leave us. Three things, stated plainly.

NAD+ boosters are not a scam. They contain what they claim, they do the biochemistry they claim, and they appear to be safe at the doses studied — including a dedicated safety trial of 1250 mg/day for four weeks in healthy adults (Fukamizu et al., Scientific Reports 2022; according to PubMed, DOI). This is a real product doing a real thing. That’s more than you can say for most of the shelf.

The thing it does has never been connected to the thing you want. Raising blood NAD+ is a surrogate for a surrogate. It stands in for cellular NAD+, which stands in for sirtuin activity, which stands in for the CR-mimicking effects, which stand in for aging more slowly. Four inferential hops, and the human data runs out after the first one. This is the same trap that ate resveratrol — GSK bought Sirtris for around $720 million in 2008 on the strength of the sirtuin story, and the lead compound was shelved by 2010. Same mechanism family, same beautiful narrative, same absence at the end.

And you can predict a trial’s result from its endpoint. That’s the finding I’d actually put money on. Across this entire literature, the softer and more effort-dependent the outcome — and the closer the funder — the more likely the benefit. Objective physiology, and it evaporates. When a supplement’s effect size tracks measurement quality that cleanly, the parsimonious reading isn’t “we haven’t found the right dose yet.” It’s that we’re mostly measuring expectation.

There’s a regulatory footnote worth knowing, too, and it’s a tidy illustration of how little any of this hinges on whether the stuff works. In late 2022 the US FDA ruled that NMN couldn’t be sold as a dietary supplement at all, on the technicality that it had been authorised for investigation as a drug first. Then in September 2025 it reversed itself — not because new evidence turned up, but because it accepted that NMN had been marketed as a supplement back in 2017, before the drug investigation started. So it’s lawful in the US again. Note what moved and what didn’t: the legal status flipped twice on paperwork about who got to market first, while the outcome evidence stayed exactly where it was.

What to do instead

Not “nothing.” Just — proportionately.

If you’re already taking it and it’s not straining your budget, carry on. It looks safe at the doses and durations anyone has actually studied — and note that ceiling, because the longest trial ran 24 weeks, which makes long-term safety unstudied rather than established. It’s doing measurable biochemistry, and I’m not going to pretend I have a trial showing you’ll regret it. Just don’t count it as a longevity intervention on your list, because it isn’t one yet.

If you’re deciding whether to start, spend the money elsewhere first. Creatine has a far larger, largely independent human evidence base with functional outcomes attached, at roughly a tenth of the price per month. Omega-3 has hard cardiovascular endpoint trials, even where they’re mixed. Both clear a bar NMN hasn’t approached.

And keep the ladder. It generalises well past this one molecule. Next time a supplement’s evidence page impresses you, ask what the primary endpoint actually was, and who paid. If the wins are all on questionnaires and walk tests while the clamps and biopsies come back quiet, you’re looking at a compound that works in the places where working is easy.

The uncomfortable summary: NMN is the best-selling longevity supplement, and it has the thinnest outcome evidence of the serious contenders. Those two facts are related, and not in a flattering way. Ranked by healthspan per hour and per pound, it doesn’t make the page.

Genuinely, though — check back in five years. The mouse data is good enough that a proper long trial with hard endpoints is worth running, and one day someone will run it. Until then it’s a promising molecule being sold as a finished answer, which is a bit like being handed the bill halfway through the meal.