NMN vs NR: What the 2026 Head-to-Head Study Found

TL;DR: A 2026 crossover trial from the University of Bergen (Berven et al., iScience) directly compared nicotinamide riboside (NR) and nicotinamide mononucleotide (NMN) at the same 1,200 mg/day dose in the same people. Industry-affiliated coverage of that trial reports NR raised blood NAD+ by roughly 161% versus about 67-69% for NMN, with the NMN arm reportedly stopped early. We could not independently confirm those exact figures from the paywalled primary text, so treat the percentages as reported, not settled, while the underlying biochemistry gives a plausible reason NR might win anyway.

Two precursors, one destination

NMN and NR are both marketed as ways to raise NAD+, the coenzyme every cell needs for energy metabolism and DNA-repair signaling that our primer on what NAD+ is covers in more depth. Neither molecule is NAD+ itself; both have to be converted inside cells, and the conversion path is where the two precursors actually differ.

NR’s route: NR is absorbed, then phosphorylated by the enzymes NRK1 and NRK2 into NMN, which is then joined to ATP by NMNAT enzymes to form NAD+. Two enzymatic steps, starting from a small, uncharged nucleoside.

NMN’s route: NMN is, on paper, already one step further down that same path. It only needs NMNAT to become NAD+. The catch is that NMN is a larger, phosphorylated molecule, and phosphate groups do not cross cell membranes easily. Some data suggest a meaningful share of oral NMN gets dephosphorylated back down to NR (or further to nicotinamide) in the gut before it is ever taken up, meaning the “shorter path” only matters if the intact molecule actually gets inside the cell in the first place.

That distinction, being structurally closer to NAD+ while being logistically farther from getting inside a cell, is the crux of the entire NMN vs NR debate, and it is worth holding in mind before looking at the headline study.

The 2026 head-to-head study, and what we could verify

The study driving most of the current “NR beats NMN” chatter is real. We located it independently: Berven et al., “The NAD-brain pharmacokinetic study of NAD augmentation in blood and brain using oral precursor supplementation,” published in iScience (Cell Press) on January 27, 2026, DOI 10.1016/j.isci.2026.114764, from a Parkinson’s-research group at the University of Bergen and Haukeland University Hospital, Norway.

Here is what the peer-reviewed PubMed abstract itself confirms: this was a phase I pharmacokinetic trial in 6 healthy adults and 6 people with Parkinson’s disease, who received 1,200 mg/day of either NR or NMN (600 mg twice daily). Blood NAD+ rose slowly and plateaued after roughly two weeks of dosing; cerebral NAD+ became measurably elevated after about four weeks; NAD-related metabolites moved faster than NAD+ itself; and responses varied a lot between individuals but were not tied to disease status or sex.

What the abstract does not state is the specific percentage comparison now circulating: that NR raised blood NAD+ by about 161% versus roughly 67-69% for NMN over the first 8 days, or that the NMN arm was discontinued mid-study. Those numbers come from secondary coverage on nad.com and nmn.com, both properties tied to ChromaDex’s Niagen NR business, the exact ingredient the story favors. The full text sits behind a Cell Press and ScienceDirect paywall we could not get past, so we are not able to independently confirm those percentages or the discontinuation claim from the primary source ourselves.

With that caveat placed up front, here is how the reported numbers compare, labeled for exactly what they are.

Why “one step closer” hasn’t translated to “wins the trial”

If NMN needs one less conversion step, the reported result is counterintuitive, until you remember that raising NAD+ is not just about enzyme count. It is about how much of an oral dose survives digestion, crosses the gut wall, and gets inside the right cells in usable form.

A 2018 Nature Metabolism paper proposed a dedicated transporter, Slc12a8, that lets intact NMN cross cell membranes directly in mice. That would remove the absorption penalty for NMN’s larger size. But a follow-up paper the next year reported it could not replicate the transport function, the original group published a rebuttal defending its methods, and the dispute has never been cleanly resolved. Independent confirmation that Slc12a8 meaningfully transports NMN in human tissue, at oral doses, is thin. In practice, a fair amount of oral NMN is thought to be dephosphorylated back to NR (via enzymes like CD73) before uptake anyway, which would mean both precursors often end up entering the same NRK1/NRK2 doorway regardless of which one you swallowed.

So the mechanistic story is not “NMN should obviously lose.” It is that NMN’s one-step advantage only pays off if the intact molecule reaches the inside of a cell, and that step is exactly where its extra size and charge work against it. A molecule that is theoretically closer to the finish line is not the same as a molecule that reliably gets to the starting gate.

The kinetics both precursors share

One thing the Bergen abstract does establish clearly, and this applies to both precursors, not just NR, is that NAD+ elevation is slow and cumulative, not immediate.

That slow-and-cumulative pattern is consistent with the older, independently verifiable trials: Martens et al. (2018) needed six weeks of 1,000 mg/day NR to show a roughly 60% rise in blood NAD+ in healthy older adults, and Yoshino et al. (2021) ran 250 mg/day of NMN for a full 10 weeks in 25 prediabetic postmenopausal women before reporting a roughly 25% improvement in muscle insulin sensitivity, alongside increased NAD+-related metabolites in muscle. Neither precursor is a same-day fix, and the Yoshino result specifically does not generalize to healthy young adults chasing longevity, since the study population was middle-aged, prediabetic, and female.

The first-pass problem both precursors share

There is a wrinkle that the “NR versus NMN” framing tends to skip: a large fraction of either molecule never arrives at your tissues intact.

Oral NR and NMN are both extensively broken down before they reach systemic circulation. Gut microbiota and the liver cleave them toward nicotinamide, which is then re-salvaged into NAD+ through the same NAMPT-dependent pathway your body already runs on nicotinamide from food. Isotope-tracing work in mice showed that orally administered NR is largely hydrolysed to nicotinamide in the gut and liver, with the intact molecule contributing far less to peripheral tissue NAD+ than the naive “you swallow NR, cells get NR” picture suggests.

That matters for interpretation. When a trial reports a blood NAD+ rise, it is measuring the endpoint both molecules converge on, not proof that one delivered more intact precursor to muscle, brain, or heart. It also explains the slow, cumulative kinetics seen across trials: you are topping up a salvage pathway, not injecting the product.

The honest summary of the biochemistry is that neither molecule has a clean, uncontested mechanistic advantage in humans. NMN is one enzymatic step closer but larger and phosphorylated; NR is smaller and membrane-permeable but needs an extra step. Both are heavily first-passed. The argument does not resolve on a whiteboard, which is exactly why the head-to-head trial data matters, and why the quality of that data matters even more.

So, NMN or NR?

If you weight only the reported 2026 crossover numbers, NR looks like the clear winner on raw NAD+ elevation, but those percentages rest on industry-affiliated retelling of a paywalled paper. If you weight the older, fully accessible, peer-reviewed trials, both precursors have independently demonstrated they raise NAD+ or NAD+-linked outcomes in humans, at different doses, timelines, and populations. The “NMN should win, it’s one step closer” argument has a real counter in the contested Slc12a8 transporter story, NMN’s larger phosphorylated structure, and the first-pass metabolism both share.

None of this overlaps with the separate question of oral precursors versus injected NAD+, which we cover in detail, including dosing ranges and the flushing/rate mechanism, in our piece on NAD+ injections vs NMN/NR dosing. And if cellular energy is the broader interest driving the NAD+ research, it is worth understanding how it connects to the mitochondrial peptides MOTS-c and SS-31, which target the same energy machinery from a different angle. Whichever precursor you are evaluating, our peptide quality and safety guide is the right starting point for sourcing questions before purity or provenance ever come up.

Frequently asked questions

Does NR really raise NAD+ more than NMN?

A 2026 crossover trial reportedly found NR raised blood NAD+ by about 161% versus roughly 67-69% for NMN at the same 1,200 mg/day dose, but those exact figures come from NR-industry-affiliated secondary coverage, not the paper’s own peer-reviewed abstract, and we could not verify them against the paywalled full text ourselves. The trial’s design and general findings, including that NAD+ rises slowly over weeks, are independently confirmed on PubMed.

If NMN is one step closer to NAD+, why wouldn’t it win?

Because enzyme steps are not the only bottleneck. NMN is a larger, phosphorylated molecule that does not cross cell membranes as easily as NR, and the transporter proposed to let it in directly, Slc12a8, has a contested, unresolved replication history. A chemically shorter path only helps if the intact molecule reaches the inside of the cell, which is exactly where NMN’s structure works against it.

Was the NMN arm of the 2026 study actually stopped early?

That specific claim appears in NR-industry-affiliated coverage of the Bergen trial, not in the study’s own peer-reviewed abstract, which does not mention discontinuation. We were not able to access the primary full text to confirm or deny it, so it should be treated as reported rather than established until an independent source verifies it.

Does taking more NAD+ precursor raise NAD+ further?

Not proportionally. Blood NAD+ rises toward a plateau over weeks rather than climbing linearly with dose, and both precursors are heavily metabolised to nicotinamide before reaching tissue. The salvage pathway they feed is enzyme-limited, so past a point you are supplying substrate to a step that is already saturated.

Does the Yoshino 2021 NMN study apply to healthy adults trying to boost NAD+ for longevity?

Not directly. That trial studied 250 mg/day of NMN for 10 weeks in 25 prediabetic, postmenopausal women and measured muscle insulin sensitivity, not general longevity outcomes in healthy people. It is solid evidence that NMN engages human metabolism, but it does not establish effects in a healthy, younger population.

Should I switch from NMN to NR based on this study?

That is a personal call weighed against the caveats above: the headline percentage gap is reported, not independently verified by us, while NR does have older RCTs (like Martens 2018) with fully accessible data behind it. Neither molecule has a large, independently confirmed trial proving clinical superiority over the other yet.

References

1. Berven H, Svensen M, et al. The NAD-brain pharmacokinetic study of NAD augmentation in blood and brain using oral precursor supplementation. iScience. 2026;29(3):114764. DOI: 10.1016/j.isci.2026.114764.
2. Martens CR, Denman BA, Mazzo MR, et al. Chronic nicotinamide riboside supplementation is well-tolerated and elevates NAD+ in healthy middle-aged and older adults. Nature Communications. 2018;9(1):1286.
3. Yoshino M, Yoshino J, Kayser BD, et al. Nicotinamide mononucleotide increases muscle insulin sensitivity in prediabetic women. Science. 2021;372(6547):1224-1229.
4. Grozio A, Mills KF, Yoshino J, et al. Slc12a8 is a nicotinamide mononucleotide transporter. Nature Metabolism. 2019;1:47-57 (2018 online).
5. Kropotov A, et al. Absence of evidence that Slc12a8 encodes a nicotinamide mononucleotide transporter. Nature Metabolism. 2020;2:461-462.

This article is educational reference material for research-use-only contexts and is not medical advice or a human-dosing recommendation. Where primary data could not be independently verified, that is stated explicitly above rather than presented as settled fact.