NMN and Blood Sugar: What the Glucose Trials Actually Found

The pooled result is a flat line

Across 15 randomised trials, oral NMN shifted fasting glucose by 0.73 mg/dL against placebo. In UK units that is 0.04 mmol/L. Your meter cannot see it. Neither can your GP's lab.

HbA1c came out at 0.03 percentage points, in the wrong direction, with a confidence interval straddling zero. Total cholesterol, LDL, HDL and triglycerides: nothing. Body weight: 0.32 kg, which is a large cup of tea.

That is the 2026 systematic review in Nutrients, pooling every parallel-group trial published to May 2026. An earlier review in Current Diabetes Reports pooled eight trials and 342 adults and landed in the same place. Its conclusion was unusually blunt for a journal: the findings do not support NMN supplementation in the general population to improve glucose and lipid metabolism.

So if you are buying NMN to move a fasting glucose reading or an HbA1c, start from that. The trials have looked, repeatedly, and found a flat line.

The one trial that found something

In 2021, Yoshino and colleagues published a 10-week trial in Science. Twenty-five postmenopausal women with prediabetes, all overweight or obese, took 250 mg of NMN daily or placebo.

They measured insulin sensitivity with a hyperinsulinaemic-euglycaemic clamp. That is the reference method: insulin is infused at a fixed rate, glucose is infused to hold blood sugar steady, and the glucose you have to give is a direct read on how well the tissues take it up.

Insulin-stimulated glucose disposal rose in the NMN arm. It did not change on placebo. Muscle biopsies showed higher AKT and mTOR phosphorylation, which is the insulin signal arriving inside the cell rather than stalling at the door.

That is a real finding in a well-run trial. It is also 25 people, and it attracted a formal Comment in Science questioning the analysis. One trial of 25 is where a question starts, not where it ends.

Why the muscle moved and the blood panel did not

These two results are not in conflict. They measure different things.

A clamp measures how much glucose skeletal muscle will take up when insulin is high. Fasting glucose measures a steady state that your liver defends tightly overnight. HbA1c averages your glucose over roughly the previous eight to twelve weeks. A modest change in muscle uptake in 25 women will not shift either.

There is a second reason the pooled numbers are flat. Most NMN trials recruited people with normal baseline glucose. You cannot improve a number that is already where it should be. Only one of the eight trials in the 2025 review enrolled a group with prediabetes — and that was Yoshino's.

So the honest summary is narrow. In healthy adults, NMN does nothing measurable to glucose control. In people with early metabolic risk, it has been properly tested exactly once.

HOMA-IR keeps leaning the same way

HOMA-IR estimates insulin resistance from fasting glucose and fasting insulin together. It is cruder than a clamp and cheaper by a factor of about fifty.

Two separate meta-analyses found it leaning the same direction. The 2025 review pooled three trials and found a reduction of 0.27 (95% CI −0.01 to 0.55, p = 0.06). The 2026 review pooled five comparisons in 160 people and found 0.22 (95% CI −0.55 to 0.11, p = 0.19), with zero heterogeneity.

Neither crossed the significance line. And the first one dissolved on inspection: excluding one 2022 trial killed it, because that trial's difference came from the placebo group getting worse, not the NMN group getting better.

Two independent pools pointing the same way is worth a real trial in people who actually have insulin resistance. It is not worth £30 a month on its own.

One oddity is worth flagging. The 2026 review found a small but robust drop in diastolic blood pressure: 2.43 mmHg (95% CI −4.21 to −0.66, p = 0.007) across 11 comparisons in 339 people. It survived leave-one-out testing. Nobody markets NMN for that, which is usually a hint that it was not fished for.

Nicotinamide riboside ran the same experiment and failed

If NMN worked through NAD+ repletion, its sister molecule should behave similarly. It does — including the null results.

Dollerup and colleagues randomised 40 sedentary men with obesity, aged 40 to 70, to 2,000 mg a day of nicotinamide riboside or placebo for 12 weeks. They used a clamp, tritiated glucose tracers, DXA and MR spectroscopy. Insulin sensitivity did not improve. Neither did endogenous glucose production, glucose disposal, glucose oxidation, lipolysis or body composition.

That is a well-powered, well-instrumented, expensive null. It was safe, and it did nothing.

Part of the explanation may be pharmacokinetic. Tracer work by Liu and colleagues showed that orally administered NMN and NR are largely broken down to nicotinamide in the liver before reaching the circulation intact. Which precursor arrives where turns out to matter less than how much of either survives the first pass.

Dose does not rescue it

The obvious objection is that the doses were too low. Mouse studies that showed metabolic benefit used up to 300 mg per kg of body weight daily. Scaled to a 75 kg human, that is 22 grams a day. Nobody is taking that.

But the human dose range has been tested, and it is flat. The 2026 review ran a dose subgroup analysis across 250 to 2,000 mg a day. No significant difference between subgroups (p = 0.93). Duration made no difference either (p = 0.50).

There is also a reason not to push the dose yourself. Oral NMN is substantially converted to nicotinamide, and the tolerable upper intake for nicotinamide sits at 900 mg a day. Ferrell and colleagues linked a terminal niacin metabolite to vascular inflammation and higher cardiovascular risk in 2024. More is not obviously better. Our dosage guide sticks to 250–500 mg for that reason.

What actually moves the number

The Diabetes Prevention Programme randomised 3,234 adults at high risk and followed them for a mean of 2.8 years. In the intensive lifestyle arm, 58% fewer people went on to develop type 2 diabetes than in the placebo arm (95% CI 48 to 66). Metformin managed 31% (95% CI 17 to 43). The lifestyle arm beat the drug, decisively.

The target was not heroic: 7% body weight loss and 150 minutes a week of moderate activity. If you are already training four or five times a week, you have cleared that bar years ago, and the remaining levers are different ones.

Muscle mass is the big one — skeletal muscle is where most insulin-stimulated glucose goes, so more of it is more storage capacity. Then there is what happens between sessions. Two weeks of reduced stepping cost ten healthy men 17% of their insulin sensitivity, which is a bigger swing than any NMN trial has produced. Sleep and alcohol both move fasting insulin faster than any capsule.

In the UK, non-diabetic hyperglycaemia is defined as HbA1c 42–47 mmol/mol, or fasting plasma glucose 5.5–6.9 mmol/L. If you are in that band, the NHS Diabetes Prevention Programme referral is free and has population data behind it.

What we would tell you at the counter

NMN is not a blood sugar product, and we would rather say so than let you find out at your next blood test.

What oral NMN does reliably is raise blood NAD+, dose-dependently, across trial after trial. That is the claim the evidence carries. The glucose story is one 25-person trial, two non-significant HOMA-IR pools, and a lot of flat lines. Anyone selling you NMN on the strength of the Science paper is quoting the exception and hiding the rule — and small trials are exactly where that trick works.

If your HbA1c has drifted into the 42–47 band, start with the intervention that has 3,234 people behind it, not the one with 25. Ask your GP for a fasting insulin alongside the HbA1c if they will run it; the pair tells you more than either alone.

And if you take NMN with trans-resveratrol anyway, take it for NAD+. Keep the dose in the range the human trials have used, keep training, and judge it on something other than a glucose meter.

References

1. Yang W, Huang J, et al. (2026). "Safety and Metabolism-Related Outcomes of Oral Nicotinamide Mononucleotide Supplementation in Adults: A Systematic Review and Meta-Analysis." Nutrients, 18(14), 2251. doi:10.3390/nu18142251.
2. Chen F, Zhou D, Kong AP, et al. (2025). "Effects of Nicotinamide Mononucleotide on Glucose and Lipid Metabolism in Adults: A Systematic Review and Meta-analysis of Randomised Controlled Trials." Current Diabetes Reports, 25(1), 4. doi:10.1007/s11892-024-01557-z. PMID 39531138.
3. Yoshino M, et al. (2021). "Nicotinamide mononucleotide increases muscle insulin sensitivity in prediabetic women." Science, 372(6547), 1224-1229. doi:10.1126/science.abe9985. PMID 33888596.
4. Yi L, et al. (2023). "The efficacy and safety of beta-nicotinamide mononucleotide (NMN) supplementation in healthy middle-aged adults: a randomized, multicenter, double-blind, placebo-controlled, parallel-group, dose-dependent clinical trial." GeroScience, 45(1), 29-43. doi:10.1007/s11357-022-00705-1.
5. Dollerup OL, Christensen B, Svart M, et al. (2018). "A randomized placebo-controlled clinical trial of nicotinamide riboside in obese men: safety, insulin-sensitivity, and lipid-mobilizing effects." American Journal of Clinical Nutrition, 108(2), 343-353. doi:10.1093/ajcn/nqy132. PMID 29992272.
6. Diabetes Prevention Program Research Group (2002). "Reduction in the Incidence of Type 2 Diabetes with Lifestyle Intervention or Metformin." New England Journal of Medicine, 346(6), 393-403. doi:10.1056/NEJMoa012512.
7. Liu L, et al. (2018). "Quantitative Analysis of NAD Synthesis-Breakdown Fluxes." Cell Metabolism, 27(5), 1067-1080. doi:10.1016/j.cmet.2018.03.018.
8. Ferrell M, et al. (2024). "A terminal metabolite of niacin promotes vascular inflammation and contributes to cardiovascular disease risk." Nature Medicine, 30(2), 424-434. doi:10.1038/s41591-023-02793-8.
9. NHS England / Public Health England (2015). "NHS Diabetes Prevention Programme: Non-diabetic hyperglycaemia." Defines non-diabetic hyperglycaemia as HbA1c 42-47 mmol/mol or fasting plasma glucose 5.5-6.9 mmol/L.

NMN for NAD+, not for your glucose panel

We sell NMN with trans-resveratrol because oral NMN raises blood NAD+, and we say plainly what the glucose trials found: nothing on fasting glucose, nothing on HbA1c, and a lean on HOMA-IR that has not reached significance. Buy it for the claim the evidence supports.

See NMN + Resveratrol →