Limitations of Current NMN Research

What this page does

This page states plainly what's weak, incomplete, or unresolved about the current human NMN evidence base. Every factual claim below is sourced directly to the primary trial or systematic review that supports it, or to an authoritative methodology reference where the point is a general research-design principle rather than an NMN-specific finding. Naming limitations honestly is not the same as dismissing the field, and this page is equally careful not to overstate the weaknesses. The fair summary, stated up front: the field is real, but the evidence is still immature.

Limitations drawn from the human trial literature

Small sample sizes. Randomized trial sizes in the published literature range from 14 participants[1] to 80 participants[2], with most falling in the 20–60 range — small enough that a single trial is often underpowered to detect anything but a fairly large effect, and vulnerable to chance findings in either direction.

Short durations. The longest published trial duration identified is 24 weeks, in a 14-participant trial.[1] Most trials run 4 to 12 weeks;[3][4] the shortest safety and biomarker trials run 14 days.[7][14] No published human trial has run long enough to speak to effects that might take months or years to develop or reverse.

Population heterogeneity. Trial populations range from postmenopausal women with prediabetes[8] to older men with diagnosed diabetes and impaired physical performance[1] to generally healthy adults in their 30s to 60s.[7] A finding in one narrow population — such as the muscle-insulin-sensitivity result specific to prediabetic postmenopausal women[8] — should not be generalized to other groups without direct testing.

Dose and formulation heterogeneity. Studied doses span roughly an eightfold range, from 250 mg/day[8] to 2,000 mg/day,[9] using both generic NMN and at least two distinct proprietary formulations (MIB-626[9] and Uthever[10]). No trial has directly compared a full range of doses or formulations against each other in the same population over the same duration, so results from one dose or formulation cannot be assumed to generalize to another.

Outcome heterogeneity and primary-vs-secondary confusion. "Physical performance" alone has been operationalized differently across trials — as grip strength and walking speed,[1] as submaximal oxygen-uptake measures distinct from VO2max,[11] as 6-minute walk distance,[2] and as muscle strength, fatigability, and stair-climbing power[9] — which complicates direct comparison across studies. Several of the more interesting positive findings were secondary or exploratory outcomes in trials whose prespecified primary endpoint was null: one trial's primary stepping-test outcome was not significant, while its secondary sleep-quality (PSQI) measure was.[3] Treating a secondary finding as if it were a trial's main result overstates the evidence.

Lack of replication. Several specific positive findings currently rest on a single trial each, without independent replication by another research group: the muscle-insulin-sensitivity result in prediabetic postmenopausal women,[8] the PSQI secondary sleep-quality finding,[3] and the small, open-label hypertension pilot's blood-pressure reduction.[12]

Industry funding and incomplete conflict-of-interest verification. Several trials in the literature disclose funding from, or author employment by, an NMN manufacturer or ingredient supplier, including Mitsubishi Corporation Life Sciences,[5][4][6] Meiji Holdings,[3] Effepharm,[10] Abinopharm and Aba Chemicals,[2] and Metro International Biotech, whose senior co-author is also disclosed as a consultant and equity owner in that company.[9] Other trials report no competing interests,[13][11] and several could not have their funding/COI status independently verified from the paper's own record.[1][14][7][12] Industry funding does not by itself invalidate a finding, but it is relevant context, and this hub reports it per-paper rather than assuming it either way.

Limitations visible in the pooled (meta-analysis) literature

Pooled analyses add their own documented limitations rather than resolving individual trials' weaknesses. A 2025 meta-analysis of 12 studies and 513 participants found NAD+ rose significantly overall but most clinically relevant metabolic outcomes did not, and its own authors rated 7 of the 12 included studies as having "some concerns" and 5 as "high risk of bias."[15] A 2024 review of 10 studies, despite a title referencing "improved" physical performance parameters, found pooled grip-strength and skeletal-muscle-index changes were not statistically significant.[16] A 2025 meta-analysis of NMN and NR trials in adults over 60 concluded current evidence does not support either compound for preserving muscle mass and function.[17] A 2026 blood-pressure meta-analysis found a modest, significant pooled effect but explicitly labeled its own evidence "preliminary and suggestive," calling for larger, longer, higher-quality trials.[18]

Broader methodological limitations, not specific to any one trial

Surrogate endpoints and limited hard clinical outcomes. Nearly every outcome measured across this literature — blood NAD+, a blood pressure number, a strength test — is a surrogate or intermediate marker rather than a hard clinical outcome like a diagnosed disease, a hospitalization, or a death. Methodological research on surrogate endpoints in clinical trials generally has repeatedly found that a treatment's effect on a surrogate marker does not reliably predict its effect on the clinical outcome that marker is meant to stand in for.[19] A surrogate marker moving in a favorable direction does not guarantee a corresponding clinical benefit.

Limited disease-specific evidence. Almost all trials cited above enroll generally healthy or narrowly-defined populations rather than people with a specific diagnosed disease NMN is being marketed to help with;[8][7] disease-specific efficacy has not been established.

Absence of long-term safety data. The longest published human trial identified runs 24 weeks;[1] multi-year safety data for continuous NMN use does not currently exist.

Absence of human lifespan data. None of the trials cited above, or identified elsewhere in this hub's review of the published literature, measured mortality or a validated aging-clock endpoint; as covered on NMN and Aging, no study has tested, or could feasibly test on any near-term timescale, an effect on human lifespan.

Publication bias and selective outcome reporting risk. As in most areas of clinical research, systematic reviews of the broader literature have found that studies with statistically significant or positive results are more likely to be published, and more likely to be published promptly, than studies with null results — and that trials reporting multiple outcomes create room for selectively emphasizing whichever outcome reached significance.[20] This hub's own trial review specifically sought out and retained null and negative results[1][13] to counter this pattern, but cannot rule out unpublished null NMN trials that were never submitted at all.

Multiple comparisons. Trials measuring several outcomes at once increase the chance that at least one reaches statistical significance by chance alone; not every trial in the literature reports having corrected its statistical analysis for this.

Inconsistent endpoint definitions. As shown above, "physical performance" alone is measured with at least four different specific instruments across the literature reviewed for this hub,[1][11][2][9] which complicates direct comparison and pooling.

What's genuinely improved, stated fairly

None of the above should read as dismissal. Compared to several years ago, multiple independent randomized, double-blind, placebo-controlled trials of oral NMN now exist rather than a handful of small pilot studies.[8][3][2][9] NAD+ and related biomarkers have been measured repeatedly and consistently across independent research groups and countries, with a reproducible dose-dependent signal.[2] Several distinct doses, from 250 mg/day up to 2,000 mg/day, have now been studied for safety,[8][6][9] and short-term safety data are genuinely accumulating, with no serious adverse events reported across any trial identified for this hub. The field has moved meaningfully past pure preclinical speculation — it just hasn't yet accumulated the volume, duration, and replication of evidence that would let this hub make stronger claims than it currently does.

Related reading

For the complete trial-by-trial breakdown, see NMN Human Clinical Trials. For safety specifically, see NMN Safety and Side Effects. For dosing actually studied, see NMN Dosage. For why animal findings don't resolve these gaps, see Preclinical vs. Human NMN Evidence.