Does Urolithin A Actually Work?

Most supplements are things you could in principle eat. Urolithin A is not. It appears in no food. What food contains are ellagitannins and ellagic acid — abundant in pomegranates, walnuts, strawberries and raspberries — and what turns those into urolithin A is a specific population of gut bacteria from the Eggerthellaceae family.^[11] The compound in the capsule is the metabolite, chemically synthesised, skipping the bacteria entirely.

That detail is the whole commercial argument, and it is also where the honest version of this story starts. If everybody's gut made urolithin A efficiently from a handful of walnuts, there would be little reason to buy it. The marketing says most people's doesn't. The research literature says it depends almost entirely on how you measure.

Three serious attempts to count converters give three different answers. The largest, a Spanish cohort of 839 people aged 5 to 90, classified about 10% as UM-0 — genuinely unable to produce any urolithin — and found that proportion stayed roughly constant across the whole lifespan.^[10][11] What changed with age was the type of production: among producers, the share making only urolithin A fell from about 85% in childhood to about 55% by the late thirties, while the share also making urolithin B rose from 15% to 45%, and then held steady.^[11] A manufacturer-run study of 100 healthy adults took a different approach — give everyone a glass of pomegranate juice and see who responds — and found only 12% had detectable urolithin A at baseline and roughly 40% converted significantly after the juice.^[9] And a 2026 crossover trial of 78 postmenopausal women, using the standard three-day ellagitannin protocol, found no non-producers at all; its authors said plainly that this reflects the limited sensitivity of that protocol rather than a true absence.^[12]

So the figure you see quoted — "only 30–40% of people convert" — is not wrong so much as protocol-dependent. A single glass of juice over a short window is a test of how briskly you convert. A three-day loading protocol is a test of whether you convert at all. Those are different questions, and they get different numbers.

The paper that launched the field is a 2016 Nature Medicine study showing urolithin A induces mitophagy both in cells and in living animals after oral intake.^[1] In C. elegans it prevented the age-related accumulation of damaged mitochondria, extended lifespan, and preserved mobility and pharyngeal pumping into old age. In rodents it improved exercise capacity in two mouse models of age-related muscle decline and in young rats.^[1]

This is a genuinely good piece of biology, and the mechanism it describes — PINK1/Parkin-mediated clearance of faulty mitochondria — is one of the few ageing pathways with a plausible, well-mapped route from molecule to muscle. It is also worth saying what it is not. A nematode that lives longer is not a human who lives longer; the gap between those two statements has swallowed most of the longevity field's promises. And the 2016 paper already had authors from Amazentis, the company that would go on to sell the compound as Mitopure — which is not a scandal, it is how translational science usually works, but it means the independent-replication clock started at zero.

The first-in-human study, published in Nature Metabolism in 2019, gave healthy sedentary older adults single doses up to 2,000 mg and daily doses of 250–1,000 mg for four weeks.^[2] Its primary outcome was safety, which it met. Its secondary findings were that urolithin A was bioavailable in plasma at every dose, and that 500 mg and 1,000 mg modulated plasma acylcarnitines and skeletal-muscle mitochondrial gene expression.^[2] That is a molecular signature. Nobody got stronger, and the study was not designed to find out whether they would.

Then came two four-month randomised trials, both manufacturer-sponsored, both of which missed their pre-specified primary endpoints.

In 66 adults aged 65–90, 1,000 mg daily for four months did not significantly improve six-minute walk distance or maximal muscle ATP production — the two co-primary outcomes.^[3] What did improve was muscle endurance: contractions to fatigue rose by 95.3 in the hand muscle and 41.4 in the leg muscle, against 11.6 and 5.7 on placebo.^[3] Plasma acylcarnitines, ceramides and C-reactive protein all fell. The authors' own conclusion was careful: urolithin A "may counteract age-associated muscle decline; however, future work is needed to confirm this finding."^[3]

In 88 middle-aged, overweight, inactive adults, 500 mg or 1,000 mg daily for four months likewise failed on its primary endpoint, peak power output, where both doses produced non-significant within-group gains of around 4%.^[4] The headline that travelled instead was muscle strength. Hamstring average peak torque rose 9.8% on urolithin A while falling 9.8% on placebo — a between-group difference of 19.6 percentage points (95% CI 2.0 to 37.2, p = 0.029).^[8] That is a real result. It is also one significant outcome out of five strength and performance measures in the same trial: quadriceps torque, hand-grip strength, VO₂peak and lean body mass all moved in the right direction without reaching significance.^[8] Six-minute walk distance rose 33.4 m in the 1,000 mg arm — clinically meaningful on its face — but that was a within-group change; against placebo it reached only p = 0.098.^[4]

Three smaller trials took urolithin A to people who already train.

Twenty resistance-trained men took 1 g daily for eight weeks in a study with no manufacturer involvement.^[5] One-rep-max bench press and squat were no different from placebo (p = 0.46 and p = 0.71). Maximal voluntary isometric contraction and repetitions to failure did improve (p = 0.048 and p = 0.011), and C-reactive protein fell relative to placebo.^[5][8] So: fatigue resistance yes, raw strength no.

Forty-two competitive male distance runners took 1,000 mg daily for four weeks during an altitude camp.^[6] Their 3,000 m time trial did not improve. Aerobic capacity showed no significant group-by-time interaction (p = 0.138). What did move was recovery — lower ratings of perceived exertion (p = 0.02) and reduced indirect markers of post-exercise muscle damage — and muscle proteomics showed upregulated mitochondrial pathways and downregulated inflammatory ones, though measured mitochondrial function itself did not change.^[6] The authors' summary is the fairest one available: urolithin A "facilitates recovery" but "did not further enhance performance in highly trained male endurance athletes."^[6]

The third is a single-blinded pilot in 20 academy soccer players, and it is a good illustration of why pilots should be read as pilots.^[7] Yo-Yo intermittent recovery distance improved by 239 m (95% CI 20 to 454, p = 0.048) and countermovement jump height by 3.3 cm (0.88 to 5.95, p = 0.020) — but countermovement jump peak force went the other way, down 136 N (−257 to −13, p = 0.046).^[8] Confidence intervals that wide, in twenty teenagers, with outcomes pointing in both directions, are hypothesis-generating and nothing more.

In 2026 a systematic review registered on PROSPERO pulled all of this together: five randomised trials, 236 participants, doses of 500–1,000 mg, durations of four weeks to four months.^[8] Outcomes were so heterogeneous that only one could be meta-analysed at all — six-minute walk distance, from two trials.

The pooled result was +17.03 metres (95% CI −5.33 to 39.40, p = 0.135), with zero statistical heterogeneity and a GRADE certainty rating of low.^[8] Everything else — strength, endurance, aerobic capacity, every biomarker — was reported narratively and labelled, in the authors' words, "exploratory signals rather than reproducible effects."^[8]

The most instructive thing in that review is a correction buried in its methods. An earlier version of the analysis had used only the 1,000 mg arm of the larger trial against the shared placebo group, and got +27.46 m (95% CI 4.54 to 50.38, p = 0.019) — statistically significant. Combining both dose arms against that shared placebo, as the Cochrane Handbook requires to avoid a unit-of-analysis error, cut the estimate by about 38% and erased the significance.^[8] No new data were collected. A statistical choice was corrected, and the field's one poolable positive finding evaporated.

For scale: the smallest change in six-minute walk distance people typically notice is around 14 m in older adults, and 30–50 m in sicker populations.^[8] The pooled estimate sits at the very bottom of that range, with an interval that includes doing nothing at all.

Three of the five randomised trials — the two four-month trials in older and middle-aged adults, and the runners' trial — were sponsored by Amazentis and include company employees among their authors.^[3][4][6] The two with no company involvement are the two smallest, twenty participants each.^[5][7]

This cuts in a direction people often don't expect. When the reviewers applied the Cochrane risk-of-bias tool, the two manufacturer trials were rated low risk of bias, and the two independent ones were rated as having some concerns — mostly for incomplete reporting of randomisation and the absence of a pre-specified analysis plan.^[8] Industry money did not make these trials sloppy. What it did was concentrate the evidence base: the large, well-run, long-duration studies all belong to one company, and the independent work is small and short.

That is why the most important item in the whole field right now is not a published paper. It is a confirmatory randomised trial of 120 healthy middle-aged adults on 500 mg or 1,000 mg against placebo, started in December 2025, with primary completion due in October 2027.^[14] Running it is the right thing to do. The fact that it is needed is also an admission that the case is not yet closed.

Partly. The manufacturer's own bioavailability study is the honest place to look: direct supplementation produced more than six times the urolithin A exposure of a glass of pomegranate juice.^[9] That is a company study making the company's argument, but the biology is not really in dispute — if you need a specific plasma level of a metabolite your bacteria may or may not make, taking the metabolite is a more reliable route than taking its precursor.

The counter-argument is just as straightforward. Nobody has ever shown that the plasma level is the thing that matters, because no trial has demonstrated a clinical outcome to attach it to. And a handful of walnuts delivers unsaturated fat, fibre, and a whole family of other polyphenols that a purified capsule does not — the same food-versus-supplement tension that runs through the broader polyphenol evidence, where a concentrated extract raised LDL cholesterol 9.5% while cutting oxidised LDL 39%, and the effect depended on which metabolites a woman's gut happened to make.^[12]

Every trial that found anything used 500 mg or 1,000 mg daily, for four weeks to four months.^[8] There is no evidence for lower doses and none for longer than four months.

The reference product, Mitopure, sells sixty softgels for US$99 — a one-month supply at 500 mg a day.^[15] The 1,000 mg dose that generated most of the trial signal therefore costs roughly $198 a month, or about $2,400 a year, indefinitely, for an effect size that a meta-analysis could not distinguish from zero.

On product quality: urolithin A has had a US FDA "no questions" letter on its GRAS notice since 2018, covering synthesised urolithin A at up to 500–1,000 mg per serving in specified foods.^[13] That is a safety determination for the ingredient, not a verdict on any product containing it. The manufacturer has publicly reported testing two cheaper Amazon products labelled as high-purity urolithin A at three separate laboratories and finding no detectable urolithin A in either — an interested party's claim about its competitors, but a reasonable prompt to buy only from a source that publishes third-party testing.

Urolithin A has been well tolerated in every trial to date, with no significant difference in adverse events versus placebo and no serious adverse events attributed to it.^[2][3] That is a genuinely clean record.

It is also a short one. The longest trial ran four months. Nothing is known about taking it for years, which is how people actually take longevity supplements. Mitophagy is a quality-control process, and the consequences of chronically nudging it upward in healthy tissue have not been studied in humans. There are no data in pregnancy or breastfeeding. And there are active trials of urolithin A in prostate cancer and in patients on immune checkpoint inhibitors — meaning its effects in cancer are an open research question, not a settled one, and anyone with a cancer diagnosis or on active treatment should raise it with their oncologist before starting.