Does Hydrogen Water Work? What 54 Human Trials and 50 Brands Actually Show
By
Jeremiah Kowalski
Human-written
Written by a human.No AI was used to write this article or create its product photography.Read our no-AI policy →
Table of contents18 sections
I’ve been drinking hydrogen water since 2023, and I test hydrogen water bottles as part of my job. I can measure whether a bottle actually puts dissolved hydrogen into water, but that still leaves the much bigger question:
Does drinking hydrogen water actually help people?
To find out, I went through the human research and separately audited 50 hydrogen-water brands to see what they tell customers and whether the science they cite really supports those claims.
My research now includes 54 confirmed human trial families. I group papers that use the same participants together, so the same 30 people do not suddenly become 90 just because researchers published three papers about them.
The answer is not a clean YES or NO. There are real positive signals in human studies, but the evidence is still too inconsistent and poorly replicated for me to say that hydrogen water broadly “works” for health or performance.
This is a long article because there is a lot to untangle, but I’ve tried to keep it concrete. If you read it through, you should not have to base your opinion on Reddit comments, YouTube testimonials or whichever hydrogen-water company has the nicest sales page. You’ll know what the human studies actually found, where the evidence is weak and which popular claims go further than the research.
The short answer
Here are the numbers I would want to know before spending money on hydrogen water:
54 confirmed human trial families are in my current research.
I identified a reported main outcome in 28 of them.
9 trial families had a clear positive main result. One additional trial had a mixed result.
Among those nine positive trial families, I identified 0 independently replicated positive findings.
Clearly interpretable daily amounts in human studies were roughly 0.5 to 2 liters per day.
Numeric hydrogen concentrations ranged roughly from 0.45 to 15 ppm, but only 10 of 54 trial families had usable numeric concentration data.
I found 0 randomized human trials directly showing that higher ppm gives a greater benefit.
A commercial connection was identified in 27 of 54 confirmed trial families.
45 of 50 brands I audited cited research.
Only 18.3% of 2,648 identifiable research citations involved humans actually drinking hydrogen-rich water.
In 42.5% of 367 study-linked brand claims, the marketing went beyond what the cited study actually found.
So yes, there is human research. There is also a rather large gap between “there is research” and “this benefit is proven.”
What health benefits does hydrogen water actually have?
This is the table I would want to see if I were deciding whether hydrogen water was worth trying.
Claim
Verdict
What the human studies actually show
Studies
Exercise performance
Mixed / not established
This is one of the better-studied areas, but results point in different directions. Some small studies reported better VO₂peak, power or performance, while others found no clear improvement.
[2][3][4]
One pre-workout dose makes you run longer
No benefit found
A randomized double-blind crossover trial in 24 trained runners found no improvement in time to exhaustion, VO₂, heart rate or post-exercise lactate.
[3]
Workout recovery / muscle fatigue
Promising, not established
Small studies have reported lower lactate, less decline in muscle function or better power endurance. The studies are small and do not give me a replicated answer yet.
[1][4]
Oxidative stress
Not established
Some studies report favorable oxidative-stress markers, while others do not. A better lab number is also not automatically the same as feeling healthier.
[5][6][7]
Inflammation
Not established
Several studies found changes in inflammatory markers or related pathways. That does not prove hydrogen water reduces inflammatory disease or makes someone feel better.
[5][6]
Blood sugar / glucose metabolism
Promising, not established
Some trials in people with diabetes, impaired glucose control or metabolic syndrome reported favorable changes. The populations and outcomes vary too much for a broad claim.
[6][8][9]
Cholesterol / lipids
Promising, not established
Several studies reported improvements in LDL, total cholesterol or HDL-related measures. I would no longer call this proven because the strongest positive findings have not been independently replicated.
[6][10][11]
Weight loss
Not established
Some body-size measures have improved, including a positive waist-circumference result, but that is not the same as showing reliable weight loss.
[6][12][13][23]
Sleep
Promising, not established
A small randomized study in adults with obesity reported better subjective sleep quality. One study is enough to make me interested, not enough to make me confident.
[12]
Memory / cognition
Not established
A one-year randomized trial in people with mild cognitive impairment found no significant overall improvement in the main cognition score. A subgroup result was more positive, but needs replication.
[14]
Parkinson’s disease
No convincing benefit
A small early pilot looked positive. A later multicenter trial with 178 participants did not find a significant benefit on its main Parkinson’s outcome.
[15][16]
Cardiovascular health
Interesting signal, not established
Studies have reported changes in endothelial function, meaning how well blood vessels respond. That is not the same as showing fewer heart attacks, strokes or other cardiovascular events.
[17]
Fatigue
Not established
Some studies reported improvements, but one dose/schedule study had no hydrogen-free control and found no difference between schedules. Long COVID evidence is also limited.
[18][1]
PMS symptoms
Promising, but mixed
One randomized study reported better PMS symptom scores, while its co-primary quality-of-life result was null. I would want another independent trial before going further.
[19]
Quality of life during radiotherapy
Promising, not established
One randomized 49-person study reported better quality of life during liver-tumor radiotherapy. It did not show better tumor response.
[20]
Preventing COVID-19 progression
No benefit found
The 675-person Hydro-COVID phase 3 trial did not reduce clinical worsening compared with placebo.
[21]
Anti-aging
Not established
One pilot study reported a borderline-positive telomere-length result. Telomeres are chromosome markers associated with aging, but changing a marker does not prove that people age more slowly.
[22]
Immune support
Not established
Some immune-cell and gene-expression measures changed in human research. That is a long way from proving fewer infections or meaningfully stronger immune protection.
[5]
The pattern is less exciting than many sales pages make it look. There are interesting narrow signals, but almost every broad claim becomes shakier once I ask whether the main result was positive, whether people actually noticed the change and whether another independent team got the same result.
How much hydrogen water should you drink?
This question comes up constantly: 500 mL? One liter? Two?
Among trials where the daily amount was clear enough to compare, researchers generally used about 0.5 to 2.0 liters per day. That is a studied range, not a recommended range.
I could easily turn that into “drink one liter per day” and give you a satisfyingly precise answer, but the science does not let me do that. No optimal daily amount has been established.
The trials used different populations, durations, hydrogen concentrations and outcomes. We do not have a proper dose-finding study showing that one daily amount works better than another.
What ppm should hydrogen water have?
PPM tells you the concentration of dissolved hydrogen. For very dilute hydrogen water, 1 ppm is roughly 1 milligram of H₂ per liter.
The numeric concentrations I found ranged from roughly 0.45 to 15 ppm. That looks like a useful range until you see how thin the data actually are.
Only 10 of 54 trial families had a numeric concentration I could extract, and those values were not all measured the same way. Some were measured directly, some were simply reported, and some came from manufacturer information or calculations.
In many studies, I also could not verify the exact concentration when the person actually drank the water. Hydrogen escapes from water over time, so “the device can make 5 ppm” and “the participant swallowed 5 ppm” are not automatically the same statement.
Does more ppm mean better results?
This is the finding that changed my own thinking the most. I test hydrogen bottles, and I used to give the bigger ppm number quite a lot of weight. A bottle claiming 8 ppm naturally sounds better than one claiming 3 ppm.
The human research does not currently show that 8 ppm produces a greater health benefit.
I found no randomized human trial directly assigning people to different hydrogen concentrations and showing that the higher concentration worked better. Some studies compared different amounts of water or different drinking schedules, but that answers a different question.
There is another simple problem with chasing ppm: ppm is concentration, not total dose.
If you drink 2 L at 1 ppm, that is nominally about 2 mg of H₂. Someone drinking 1 L at 2 ppm also gets about 2 mg of H₂ on paper.
That does not mean total dose is the only thing that matters. It means a single ppm number tells only part of the story, and we do not yet have human evidence establishing the best dose-response relationship.
So if I were choosing between a 3 ppm bottle and an 8 ppm bottle today, I would not pay a huge premium just because I expected 8 ppm to produce a bigger health effect.
More dissolved hydrogen is more dissolved hydrogen. More proven benefit? Not established.
Some positive results were things people could actually notice
I also wanted to know what “positive” meant. Did someone sleep better or perform better, or did one number in their blood move?
Across the positive main results I could classify, I found 10 positive components:
Type of positive result
Components
Symptoms or another change a person could notice
3
Physical or functional performance
2
Clinical outcome
2
Biomarker or body measurement
3
Mechanistic-only result
0
A biomarker is simply something measured in the body, such as a blood value, body measurement or biological marker. Biomarkers matter, but I don’t want to turn them into something they are not.
The clearest example is telomere length. Telomeres are protective structures at the ends of chromosomes and often appear in discussions about biological aging.
One hydrogen-water pilot trial reported a positive telomere result. Interesting? Absolutely. Proof that hydrogen water makes people feel younger, live longer or slows aging? No.
The person drinking the water cannot feel a telomere measurement change.
The biggest weakness I found was replication
A trial can produce an exciting result. The harder test comes next: can another independent research team run a similar experiment and get the same answer?
That is replication, and this is where the evidence becomes much less reassuring.
Among the 9 trial families with a clear positive main signal, I identified 0 independently replicated positive findings. Eight had no replication identified, while one had a related result from the same research group or network.
This does not mean those nine findings are false. It means they are still single-study signals, not benefits I would describe as firmly established.
Parkinson’s disease is a good example of why I care about this. A small 2013 pilot study reported a positive result and looked encouraging. A later multicenter study with 178 participants tested the question again and did not find a significant benefit on its main Parkinson’s outcome.
Small positive studies are worth following. They are not the finish line.
Hydrogen-water marketing is much more confident than the evidence
The other half of my investigation was the marketing.
I audited 50 hydrogen-water brands and collected 1,589 health, performance, wellness and mechanism-related claim concepts. Then I checked what kind of science sat behind them.
I separated the evidence into two simple buckets:
Direct evidence means people actually drank hydrogen-rich water and the study measured the thing being promised.
Indirect evidence means the paper measured something related, such as a biomarker, an animal outcome or a nearby biological mechanism.
Direct evidence is better, but it still does not automatically mean the claim is proven.
Here is what I found:
Claim
Brands
Direct vs. indirect evidence
My verdict after 54 trial families
Oxidative stress
50/50
14 direct / 181 indirect
Not established
Recovery
44/50
35 direct / 134 indirect
Promising, not established
Inflammation
40/50
23 direct / 99 indirect
Not established
Exercise performance
39/50
91 direct / 73 indirect
Mixed / not established
Cognition
39/50
21 direct / 47 indirect
Not established
Anti-aging
37/50
1 direct / 54 indirect
Not established
Immune support
24/50
1 direct / 20 indirect
Not established
Sleep
19/50
10 direct / 16 indirect
Promising, not established
Blood sugar / glucose
18/50
9 direct / 16 indirect
Promising, not established
Cholesterol / lipids
14/50
8 direct / 13 indirect
Promising, not established
The oxidative-stress row still jumps out at me. Every single brand I reviewed used some version of that claim, yet after looking at the human trials I still cannot give a broad antioxidant claim a confident YES.
Anti-aging is even more obvious. 37 of 50 brands made claims in that category, but the human evidence still does not show that drinking hydrogen water makes people age more slowly.
A telomere result is not the same thing as slower human aging.
A long bibliography can be very misleading
45 of 50 brands cited research. At first glance, that sounds excellent.
Then I counted what those references actually were.
Across the brand audit I recorded 2,648 identifiable research citation occurrences. Only 484, or 18.3%, involved human studies where people actually drank hydrogen-rich water.
So more than four out of five citations were something else. That can include animal research, cell research, reviews, different hydrogen delivery methods and other evidence that may be scientifically useful but does not directly answer the question a buyer probably cares about:
What happens when a person drinks hydrogen water?
I also matched individual studies to individual marketing claims. There were 367 linked brand-claim pairs, and in 156 of 367, or 42.5%, the marketing claim went further than the cited study result.
In plain English, more than 4 in 10 stretched the science.
That does not mean I think 42.5% of brands were deliberately lying. I have no idea what someone inside a company believed when writing the page.
It means the consumer-facing promise was stronger than the study behind it.
My strictest marketing test produced a zero
I pushed the audit one step further.
Did the brand’s claim directly match a beneficial main outcome that had been clearly defined before the trial?
The main outcome is basically the result the study was built around. This matters because a trial can measure 20 different things, miss its main target and still find one or two interesting positive numbers elsewhere.
Those extra findings may be worth studying. They should not automatically turn the entire experiment into proof of the marketing claim.
Under that strict standard, 0 of 367 linked brand-claim pairs passed.
Zero.
That does not mean every hydrogen-water claim is false. It means I could not verify a single brand-claim link in this sample that cleared my strongest evidence bar.
For an industry that loves the phrase “science-backed,” I think that deserves attention.
I expected a few famous studies to power the whole market. They don’t
I went into this expecting another neat finding: maybe hydrogen brands all recycle the same five or ten papers until the bibliography looks enormous.
The data did not really support that.
The 2,648 citation occurrences represented 1,352 different references, and the 20 most frequently used papers accounted for 19.79% of all citation occurrences. The literature brands cite is more spread out than I expected.
There are still obvious hubs. The most widely used individual paper in my audit was a 2007 Ohsawa study, cited by 22 of 50 brands and appearing 94 times.
Here is the catch: it was preclinical research involving cells and animals, with hydrogen delivered through routes that included inhalation. It was not a human trial showing that drinking hydrogen water produces the health benefits consumers may associate with those citations.
Widely cited does not mean directly relevant.
Who is behind the human research?
Commercial involvement was another question I kept coming back to.
In the 54 confirmed trial families:
Commercial status
Trial families
Commercial connection identified
27
No connection identified
11
Unclear
13
Unknown
3
“Commercial connection” is intentionally broad. A company might fund the research, provide the hydrogen-water product, donate equipment or have a relationship with one of the authors.
Those situations are not equivalent. Providing a bottle for a study is not the same as paying the researchers, and neither automatically makes the result wrong.
Among the 9 positive trial families, I found some form of commercial connection in 6. Explicit commercial or industry funding appeared in 4 of 9.
That is useful context, but I would not turn it into “most positive hydrogen-water studies are industry-funded.” The numbers do not support that wording.
The question I care about more is independence. And right now, the bigger problem is still that I found no independent replication among those nine positive main signals.
So, does hydrogen water work?
I would love to end this investigation with a big YES or NO. It would make this article shorter, easier to write and probably much easier to share.
The science does not let me do that.
My answer is sometimes it looks promising, but broad benefits are not proven. There are real randomized trials, placebo-controlled trials and positive findings. There are also null trials, small studies, missing dose information, one-off results and a serious replication problem.
This is also why I am suspicious of extreme certainty from either side.
If someone tells you with 100% confidence that hydrogen water works, their confidence is much stronger than the human evidence allows. After reading this research, I have a hard time believing anyone who has seriously worked through the studies could be that certain.
The same goes for the hardened skeptic saying hydrogen water is a scam. You see plenty of that in online forums too. The positive human findings are limited, but they exist. Dismissing them all is not a particularly scientific position either.
The boring answer is unfortunately the most accurate one:
We do not know yet.
Some specific benefits look promising. Some specific claims have failed in controlled trials. For most of the big health promises, the evidence sits somewhere in between.
What this changed about how I choose hydrogen bottles
I separate three questions now.
1. Does the bottle actually produce dissolved hydrogen? That can be tested with the right equipment, although home hydrogen meters have limits.
2. How much hydrogen does it produce? That matters because a weak-output bottle may not resemble the water used in human research at all.
3. Does a higher ppm number mean a bigger health benefit? This is where I changed my mind.
I used to put more weight on the biggest ppm number. I still want a bottle that produces a meaningful amount of hydrogen, but I no longer assume that 8 ppm must be healthier than 3 ppm because human studies have not shown that.
So I would judge a hydrogen bottle by its verified output, credible lab documentation, safety, build quality, warranty and whether there is a real company standing behind it. I would not pay a huge premium for ppm alone.
One limitation I don’t want to bury
My current research includes 54 confirmed trial families, but I also have 197 candidate records that remain unresolved or do not contain enough information for me to confidently include or exclude them.
That means this is not a complete census of every hydrogen-water trial ever conducted. Some of those records could eventually add new eligible trials and change percentages.
That is why I prefer saying:
“I did not identify a randomized human ppm comparison.”
rather than:
“No such study exists anywhere.”
The first sentence is what my research shows. The second would pretend I know more than I do.
Where I land
I’ve been drinking hydrogen water for years, and I still plan to keep drinking it. But I treat it as something interesting that I am experimenting with, not as a proven treatment.
The research was not all negative. There are positive studies. The problem starts when I ask the next questions.
What dose is best? We don’t know.
What ppm is best? We don’t know.
Does more ppm work better? It has not been established.
Have the positive main findings been independently repeated? Not among the nine positive trial families I reviewed.
So I end up somewhere between the two loudest opinions online. Hydrogen water is not proven magic, but I don’t think the research justifies dismissing it as nonsense either.
For now, promising is the word I am comfortable with.
Proven is not.
I’m going to keep following hydrogen-water research closely, and I’ll update this article when new human trials give us better answers. If this is a subject you care about, bookmark this page. My view is not fixed, and neither should it be if the evidence changes.
Research details and sources
Methodology – how I built the 54-family database
What this review covers
The current confirmed set contains 54 human trial families involving people who drank hydrogen-rich water (HRW). A family is a distinct participant cohort or trial; it is not necessarily one paper. Publications, registry entries, protocols, and supplements about the same participants are grouped together so repeat reporting does not inflate the trial count. The complete grouped reference list is in the 54-family source catalogue.
This is a structured evidence update, not an exhaustive systematic review or meta-analysis. We reconciled the prior candidate and trial records with targeted checks of trial identity, outcomes, hydrogen exposure, disclosures, replication, and dose comparisons. The final screened pool contains 204 candidate-family records: 6 were eligible new families, 1 was a duplicate publication, 106 lacked enough information to decide eligibility, and 91 remained unresolved. Thus 197/204 candidates remain insufficient or unresolved. “Not identified” below means not found in the reviewed records; it does not establish that something does not exist.
How trial families were assembled
We started with the retained candidate and trial-family records and checked whether each candidate represented an eligible human intervention involving oral consumption of water with dissolved hydrogen. A product name, marketing description, or planned hydrogen-generating dose alone was not treated as proof that participants consumed HRW. Where accessible, the article, registry, protocol, or supplement was used to confirm the intervention and connect reports to a cohort.
Reports were grouped when identifiers or source details showed that they described the same participant cohort or trial. We used registry identifiers, DOI/PMID and other stable identifiers, participant and arm descriptions, protocol details, investigator groups, institutions, recruitment timing, and explicit report-to-trial links. A primary paper, its registry record, protocol, preprint, and supplement can therefore appear together under one family. Related work by the same investigators was not merged automatically when the evidence indicated a separate cohort. A repeated paper or duplicate record did not create another family.
The included population is people, and the intervention is drinking HRW. Animal and cell studies, inhaled or injected hydrogen, and other non-drinking routes do not count as human drinking-water trials. Candidate records whose available sources could not establish the actual intervention or trial identity remain insufficient or unresolved rather than being forced into the confirmed cohort. The screening is limited to the retained and targeted records; it does not claim to have searched every database and language exhaustively.
How study results were classified
For each family, we recorded whether a main (primary) outcome could be identified from the report or linked registry, whether it appeared to have been specified in advance, and what the source reported. A result could not be upgraded to primary merely because it was favorable or statistically significant. If a report measured many outcomes without one clearly identified main outcome, the primary result remains unknown or not identified.
• Positive primary signal: the reviewed source reports a favorable result for an identified main outcome. This is a descriptive signal, not proof of efficacy. It does not by itself establish a between-group effect, clinical importance, a noticeable benefit, or replication; single-arm and within-arm results are not presented as controlled treatment effects.
• Null / no clear effect: the available report does not show a clear favorable result for the identified main outcome. This does not prove that every possible effect is zero.
• Mixed: co-primary components or main outcomes point in different directions. One such family reported improved PMS symptom scores (PMSS) and a null WHOQOL-BREF quality-of-life result; it is not collapsed into a wholly positive result.
• Secondary or exploratory: additional outcomes, subgroup analyses, within-arm changes, or favorable findings outside the identified main outcome are kept separate from the positive-primary count. The UMIN stool-consistency finding, for example, is not counted as a positive primary component.
• Unknown / not identified: the reviewed sources do not support a reliable classification. Missing information is retained as missing rather than treated as a negative result.
The final synthesis identifies a reported primary outcome in 28 families. Two additional families have a registry-only primary outcome but no results located. Of the 28 families with a reported primary outcome, 9 have an unambiguous positive primary signal and one more has the mixed result described above. The 9/28 figure is a descriptive fraction among families with an identified reported primary outcome, not an efficacy or “success” rate. Prespecification and between-group estimates are incomplete or unclear for some records.
What a positive outcome represents
Positive primary outcome components were classified by what they measure:
• Consumer-perceptible: a symptom or experience a participant may notice, such as symptom scores or quality of life.
• Functional: performance or ability measured through a task or function, such as exercise performance.
• Clinical: a disease-related clinical measure or response.
• Biomarker: a laboratory value or body measure that may be relevant to health but is not itself proof that a participant felt better or had a better clinical outcome.
• Mechanistic: a biological process or pathway that may help explain an effect but is not itself a participant benefit.
The final component-level count is 10 positive primary components: 3 consumer-perceptible, 2 functional, 2 clinical, 3 biomarker, 0 mechanistic, and 0 unknown. This comprises the nine unambiguous positive-signal families plus the positive PMSS component of the mixed family. Categories are descriptive and do not weight sample size, precision, clinical importance, or replication. A biomarker such as telomere length cannot be translated into a felt anti-aging benefit.
How hydrogen exposure was handled
The exposure table keeps separate fields for water volume, concentration, estimated hydrogen mass, source basis, and timing. Raw values and units are preserved alongside normalized values where a calculation is supported.
• Daily water volume: reported per-day regimens are kept separate from a one-time or exercise-day serving. Schedules stated only as a number of glasses are not converted to liters unless the glass volume is supported.
• Concentration: for dilute aqueous hydrogen, a source-supported mg/L value may be described approximately as ppm. ppb values are converted only where the unit conversion is appropriate. The numeric records span about 0.45–15 mg/L, but the underlying values have different evidence bases.
• Estimated H₂ mg/day: concentration is multiplied by daily volume only when both can be paired for that same trial and the source supports the calculation. Acute servings are described as acute amounts, not daily doses. Inconsistent source values are left unresolved.
• Measured, reported, and derived values: these are labeled separately. A paper-reported concentration, a manufacturer-supplied value, a nominal calculation from a stated mass and volume, and a direct assay are not treated as equivalent measurements.
• Generation versus consumption: a value measured or claimed when water is produced is not assumed to be the concentration when a participant drinks it. Measurement method, storage, timing, and point-of-use assay remain unknown when the source does not establish them.
• Missing values: no concentration, dose, or consumption-time value is imputed. Product specifications are not substituted for trial exposure.
Only 10/54 families have a numeric concentration field, and these records combine measured, reported, manufacturer, and nominal/derived bases. Clearly interpretable daily regimens span about 0.5–2.0 L/day in the subset where the daily volume can be compared; that is a studied range, not advice. Concentration data are incomplete and often do not verify what was present at consumption. No optimal or recommended consumer dose is established.
How concentration comparisons and dose response were assessed
We manually separated randomized comparisons of different dissolved-H₂ concentrations from studies that varied the amount of water or the drinking schedule while holding concentration constant or leaving it unreported. In the confirmed cohort and targeted comparison search, 0 randomized human ppm comparisons were identified. Two amount/schedule comparisons are separate: one varied daily HRW volume at the same reported concentration, and one varied the number of glasses across schedules without a usable trial-specific ppm comparison. Neither demonstrates that higher ppm produces more benefit or identifies an optimal amount. A single unreviewed direct ppm comparison could change this absence finding, so it is scoped to the records reviewed.
We did not fit a dose-response model or compare outcomes across studies as though concentration were randomized. Populations, conditions, durations, volumes, devices, endpoints, assay methods, and timing differ too much, and exposure data are sparse. Cross-trial patterns are not evidence that one concentration caused a different result.
How replication was assessed
Replication was evaluated for the identified positive primary signals, at the level of the finding rather than by counting papers. Linked reports from one cohort are one trial family. A result from the same investigator group or research network is labeled same-group/network replication and is not counted as independent. Independent replication requires a separate research team and a sufficiently comparable endpoint/result to count as a repeat test; a similar topic, institution, product, or related endpoint alone is not enough.
For the 9 adjudicated positive primary-signal families, the current map records 8 with no replication identified, 1 same-research-group/network result, and 0 independent replications identified. This is a finding within the mapped sources, not proof that no independent study exists anywhere. Lack of independent replication also does not prove that the original result is false.
How commercial connections were coded
We reviewed funding, acknowledgements, competing-interest statements, author affiliations, registries, and reported product or equipment support. Funding and in-kind support are separate. A product supplied for a study is a commercial connection even if the paper declares no conflict; it is not automatically cash funding. A no-conflict statement does not override a separately reported supply or funding relationship.
Each family retains one of four overall statuses: YES (a connection was identified), NO_IDENTIFIED (none was found in the reviewed sources), UNCLEAR (the available disclosures do not resolve the question), or UNKNOWN (the relevant information was unavailable). Type fields separately record industry/commercial funding, product supply or donation, device/equipment/material support, author/company relationships such as employment or other disclosed financial ties, and unclear or unknown details. These types may overlap.
For the 54 confirmed families the tally is 27 YES, 11 NO_IDENTIFIED, 13 UNCLEAR, and 3 UNKNOWN. Among the nine unambiguous positive-signal families, six had an identified connection, two had none identified, and one was unclear; four had explicit commercial/industry funding types. These descriptive counts do not show that a company connection caused a result or that a study is biased.
How the 50-brand marketing audit was counted
The audit is a purposive sample of 50 hydrogen-water brands, not a random or market-representative sample. Researchers reviewed first-party brand pages and recorded claim concepts and the research citations visible on those pages. Multiple phrasings of the same concept within a brand were normalized to a within-brand claim concept; each repeated citation appearance was retained as a citation occurrence. Stable identifiers and bibliographic details were used to merge duplicate versions of the same reference into a canonical reference. The source snapshots preserve the page/citation-level provenance.
The final audit contains 1,589 normalized relevant claim concepts, 2,648 identifiable research citation occurrences, and 1,352 canonical references. Citation occurrences are repeated mentions, not unique papers, unique claims, or independent studies. Forty-five of the 50 brands cited research. Only 484/2,648 (18.3%) research citation occurrences were classified as human drinking-HRW primary or secondary publications. This classification says what kind of source was cited; it does not establish that the cited item supports the adjacent claim.
An individual claim was linked to a cited study only where the page citation and claim association could be resolved. The analysis includes 367 linked brand-claim pairs. A claim was classified as broader than the cited result in 156/367 (42.5%) pairs. That is an evidence-scope flag, not a finding of deception or intent.
For the strictest test, a linked claim had to match a beneficial main outcome that was clearly specified before the trial. A favorable secondary result, biomarker-only result, related mechanism, or outcome that was not verified as prespecified could not pass this test. 0/367 pairs passed. Four plausible matches had unclear prespecification and were not counted as verified support. This result means the audit did not verify any pair meeting that strict rule; it does not prove that every marketed claim is false.
Direct and indirect evidence behind claims
Direct evidence for a consumer health claim requires a human study in which participants drank HRW and the study measured the outcome the claim promises. “Direct” describes relevance of the evidence, not a guaranteed positive result or proof of efficacy. Indirect evidence includes a related but different endpoint (such as a biomarker when the claim promises a felt benefit), an animal or cell experiment, a review that summarizes other work, another hydrogen delivery route, or a mechanistic result. Such work can be scientifically relevant while still failing to answer whether drinking HRW produces the promised outcome in people.
How I audited 50 hydrogen-water brands
Purposive sample: 50 brands; 1,589 normalized relevant claim concepts, 2,648 identifiable citation occurrences and 1,352 canonical references. Forty-five brands cited research. Only 484/2,648 (18.3%) occurrences were human drinking-HRW sources. Of 367 linked brand-claim pairs, 156 (42.5%) went beyond the cited result; 0/367 passed the strict beneficial, clearly prespecified primary-outcome test.
Direct evidence means human participants drank HRW and the study measured the promised outcome. Indirect evidence includes related biomarkers, animal or cell work, reviews, other hydrogen routes and mechanisms. Direct describes relevance, not proof of efficacy.
What I excluded and candidate dispositions
The confirmed cohort contains 54 distinct human trial families involving oral hydrogen-rich water. Animal and cell research, inhalation, injection, hydrogen-rich saline, baths and other non-drinking routes were excluded. Related publications about the same cohort were grouped as one family.
The candidate pool contains 204 records: 6 eligible new families, 1 duplicate, 106 insufficient-information records and 91 unresolved records. Thus 197/204 remain insufficient or unresolved. This is a structured update, not an exhaustive systematic review.
The full 54-family database
Each row represents one independent trial family; related publications and registry records are grouped. Unknown means information was not established in reviewed sources.
Primary retinitis pigmentosa 13 enrolled; 24 eyes analyzed (2 eyes excluded for no light perception)
single-arm prospective before-after clinical study NO_CONCURRENT_COMPARATOR
400–500 mL twice daily for 4 weeks
NOT_PRESPECIFIED No single prespecified primary endpoint located; visual function and ocular structure assessed Safety: Authors state no adverse impact on IOP; other adverse events not systematically characterized in abstract
NO_IDENTIFIED UNKNOWN: Authors declare no known competing financial interests or personal relationships; Key Scientific Research Project of 900th Hospital (2022ZD04); Fujian provincial grants (2020Y0076, 2020J05282); Postdoctoral Science Foundation (48678); National Natural Science Foundation of China (81870255)
NOT_PRESPECIFIED No prespecified primary endpoint identified; metabolic and DNA oxidative markers Safety: NOT_REPORTED
YES AUTHOR_EMPLOYMENT: Footnote states conflicts of interest: None; S. Sato is affiliated with Dlles In Co. Ltd.; Grant-in-Aid #0204 for AntiAging Medical Sciences, Japanese Center for AntiAging Medical Sciences
Habitual electrolyzed hydrogen water consumption >500 mL/day, at least 5 days/week for >6 months
NOT_PRESPECIFIED No trial primary endpoint; oxidative stress and blood biomarkers Safety: NOT_REPORTED
YES COMMERCIAL_FUNDING;AUTHOR_EMPLOYMENT;DEVICE_MANUFACTURER: Y. Watanabe received Nihon Trim funding for this study; S. Kabayama is employee of Trim Medical Institute Co. Ltd.; other authors report no conflicts; funders said to have no role; Nihon Trim Co., Ltd. grant 2022-0250 and RIKEN Compass to Healthy Life Research Complex Program (RC-01) of JST
BENEFIT Change in HRCT and composite physiologic index (CPI) from baseline to week 48 Safety: Reported: 7/44 vs 10/43 adverse events; P=.706
UNCLEAR UNKNOWN: Not found in extracted article text; disclosure review needed; National Key Clinical Specialty Construction Project 2021; reviewed full-text acknowledgements also identify public support; exact sponsor role not stated
Low dose: 2×330 mL HRW + 330 mL ordinary water/day; high dose: 3×330 mL HRW/day, 8 weeks
UNCLEAR Blood uric acid at weeks 4 and 8 vs baseline Safety: Authors describe feasible/safe; adverse event counts not extracted
YES PRODUCT_DONATION: Article declaration must be read from full text; not extracted in this pass; National Natural Science Foundation of China 82200508, 81770855; Shandong Youth Innovation Team 2023KJ184; Shandong First Medical University 2019QL010; Beijing Huoliqingyuan supplied H2 water
Keloid patients scheduled for two-stage operations 21 randomized; arm counts not in reviewed abstract
double-blind randomized pilot parallel trial PLACEBO
Drinking hydrogen water; preparation and concentration unavailable in reviewed abstract
UNCLEAR Keloid symptom/severity and inflammatory cytokines; exact prespecified primary not available Safety: Authors call approach safe; event counts not in abstract
UNCLEAR UNKNOWN: Authors declare no conflict of interest; Funding not available in reviewed abstract; PubMed grant list not confirmed in article
1260 mL over 15 min in three doses; H2 concentration not reported in accessible article
NO_CLEAR_EFFECT Intraocular-pressure response HRW vs hydrogen-free placebo Safety: No adverse event summary located; IOP outcome is physiological safety endpoint
NO_IDENTIFIED UNKNOWN: Authors state no conflict and no support by pharmaceutical company; Faculty of Science, Palacký University projects IGA_PrF_2019_005, IGA_PrF_2020_008, IGA_PrF_2023_004
Steroid-refractory or steroid-dependent chronic graft-versus-host disease after allogeneic HSCT 24 enrolled; 18 received 12 months; 2 lost follow-up; 4 died within 12 months
BENEFIT Objective response rate; design powered against 50% response threshold Safety: No significant toxicity observed; explicit adverse-event counts not extracted
YES MATERIAL_SUPPORT: Authors declare no commercial or financial relationships; supplier product support acknowledged; National Natural Science Foundation of China 81800180; Beijing Huoliqingyuan Beverage supplied HRW
UNKNOWN Safety and tolerability Safety: No outcome results available; NOT_REPORTED
NO_IDENTIFIED UNKNOWN: No publication disclosure; protocol/registry-only evidence; Stony Brook University listed as sponsor; protocol lists no independent grant in reviewed protocol
NO_IDENTIFIED UNKNOWN: Not reported in registry/conference abstract reviewed; University of Jordan listed as registry sponsor; grant funding statement not located
Adults with myalgic encephalomyelitis/chronic fatigue syndrome 23 randomized; fatigue endpoint completed by 11/11 H2 participants and 11/12 placebo participants
pilot randomized placebo-controlled parallel trial PLACEBO
Drinking hydrogen-enriched water, up to five glasses/day; 4-week pilot
NO_CLEAR_EFFECT Fatigue severity Safety: Reported; adverse effects led to treatment discontinuation for 27.2% in active arm
UNCLEAR UNKNOWN: No potential conflict of interest reported by authors; No funding statement located on accessible publisher abstract/full-text landing page
Chronic functional constipation 30 enrolled ITT; 28 completed and analyzed per protocol
single-arm prospective clinical intervention, per paper NO_CONCURRENT_COMPARATOR
EHARW 20 mL/kg/day in three servings; BTM-1200 generator; four weeks
BENEFIT Change in mean complete spontaneous bowel movements (CSBM) per week Safety: Authors report no adverse effects during study period
YES COMMERCIAL_FUNDING;DEVICE_DONATION: Authors declare no conflict of interest; Biontech Co., Ltd. grant 2021-51-0662; company supplied EHARW generator BTM-1200
Randomized double-blind placebo-controlled crossover Hydrogen-free control water
Exposure details not reported
NO_CLEAR_EFFECT Time to exhaustion Safety: No adverse effects that hindered participation; fullness sensation after both waters; 2 participants noticed control taste
YES MATERIAL_SUPPORT: Registry lead sponsor is University of Novi Sad; commercial funder/supplier is separately named. No sponsor no-role statement located.
BENEFIT Waist circumference (cm) Safety: One mild transient diarrhea in EXP2; increased energy in one EXP2 participant; reduced dizziness in another; evening dizziness/food cravings in one CON2; mild weakness in one CON2
YES PRODUCT_SUPPLY: Official registry lead sponsor is University of Novi Sad; paper identifies tablet supplier and affiliated coauthor. Missing funding and COI statements remain not reported.
Colorectal cancer patients receiving mFOLFOX6 chemotherapy
Randomized, single-blind, placebo-controlled parallel clinical trial Placebo water
Exposure details not reported
UNCLEAR Liver function markers after chemotherapy; no single prespecified endpoint located Safety: Chemotherapy-associated hepatic injury grades; not a general AE tally
NO_IDENTIFIED NONE_IDENTIFIED_IN_REVIEWED_SOURCES: Public grant acknowledged; no commercial product/company tie documented in reviewed article. Missing COI statement remains distinct.
Male university athletes in rugby, track and field, soccer, basketball
Randomized single-blind placebo-controlled two-period crossover exercise trial Placebo water
Exposure details not reported
UNCLEAR No single primary endpoint located; performance, HRV and metabolic outcomes reported Safety: No adverse-event results section identified in reviewed full text
YES EQUIPMENT_SUPPORT: Material support is documented and separated from cash funding. No manufacturer-employed author is identified.
Patients with chronic lung disease and baseline SpO2 90–95% 10: COPD 7, asthma 2, pulmonary tuberculosis 1
non-randomized observational pilot before-after intervention NO_CONCURRENT_COMPARATOR
Hydrogen-producing tablets in HRW for four weeks; then physical training
NOT_PRESPECIFIED Oxygen saturation before/after HRW; no prespecified primary located Safety: NOT_REPORTED
YES AUTHOR_EMPLOYMENT;PRODUCT_SUPPORT: No conflict statement not yet verified from full text; author Alex Tarnava and supplied tablet relationship require review; No external funding reported
NO_CLEAR_EFFECT ADAS-cog total score after one year Safety: NOT_REPORTED
YES PAST_ADVISOR_RELATIONSHIP: Ohta disclosed prior scientific-advisor role at Blue Mercury Inc. (2005–2008), stated no involvement in this study; JSPS Grants-in-Aid 23300257, 24651055, 26282198 (S.O.); 23500971, 25350907 (K.N.)
Healthy university students aged 21–23 Experiment II n=14 (10 men, 4 women); H2W and control groups n=7 each; separate PtW experiment I n=7 excluded from HRW family outcomes
experiment II controlled parallel product comparison; no randomization statement located CONTROL_WATER
Oral H2W 500 mL/day for 7 consecutive days
NO_CLEAR_EFFECT Blood/urine oxidative stress markers; prespecified primary not located Safety: PtW-specific ALT finding excluded because it came from separate experiment I
YES UNKNOWN: No conflict statement located in reviewed full-text PDF; No external funding statement located; H2W and control water were supplied by Bleu Mercury Co.
Randomized, double-blind, placebo-controlled parallel trial Placebo water
Exposure details not reported
MIXED_BY_TIMEPOINT Ln_RHI (all participants), measured at 1 h, 24 h and after 14 days Safety: No adverse events recorded as result of clinical study
UNCLEAR INSTITUTIONAL_SPONSOR_AFFILIATION; PRODUCT_LINK_UNCLEAR: No specific cash funding; registry says self-funded. Institute/clinic sponsorship is documented but no link to HRW manufacture or sale was identified.
BENEFIT Quality of life during radiotherapy; no registry primary located Safety: Questionnaires monitored potential adverse effects; count not extracted
NO_IDENTIFIED UNKNOWN: No conflict statement not located in extracted full text; Daimaru Research Foundation grant awarded to YG
Healthy male sports-science students, age 19.32±0.84; training ~16h/week UNKNOWN
UNKNOWN PLACEBO
UNKNOWN
See outcome rows; no overall trial classification. A priori power based on performance; no single registered primary endpoint identified Safety: UNKNOWN
YES Headrogen Health/Hydrogen Health PTY LTD test product and manufacturer concentration
Adults with impaired fasting glucose, age 35–55 UNKNOWN
UNKNOWN PLACEBO
UNKNOWN
See outcome rows; no overall trial classification. Glucose/metabolic indicators assessed; no single explicitly named primary endpoint identified Safety: UNKNOWN
See outcome rows; no overall trial classification. No single prespecified primary endpoint identified; BAP, d-ROMs, 8-OHdG, PBMC apoptosis, immune cells and RNAseq Safety: UNKNOWN
YES Commercial water sources and Coway support; HRW from Koreahydrogenwater Corp.
Healthy physically active men, age 19.4±0.3 UNKNOWN
UNKNOWN PLACEBO
UNKNOWN
See outcome rows; no overall trial classification. Exercise performance and oxidative stress markers; no registered primary endpoint identified Safety: UNKNOWN
YES Commercial hydrogen-generating kit used; product-support status unstated
Trained male track and field runners, age 17.5±1.8 UNKNOWN
UNKNOWN PLACEBO
UNKNOWN
See outcome rows; no overall trial classification. Time to exhaustion at maximal aerobic speed explicitly identified as primary aim/outcome; no preregistration located Safety: UNKNOWN
UNCLEAR Nutristamina product supplier; donation/funding/author relationship not identified
NO_SINGLE_PRIMARY_RESULT; several secondary/intervention-by-time signals, interpret cautiously Multiple appetite, body composition, sleep, quality-of-life and biomarker outcomes; no single primary endpoint identified Safety: UNKNOWN
YES Funding statement, intervention supplier statement, and disclosed employment in original paper
NO_SINGLE_PRIMARY_RESULT; most reported findings were nonsignificant trends; paper reports a BMI comparison, with clinical meaning uncertain No single prespecified primary endpoint identified; body composition and broad biochemical, inflammatory and oxidative-stress panel Safety: UNKNOWN
YES Funding statement and tablet product provision in primary paper/source review
Adults with early COVID-19 or exposed/test-positive contacts managed as outpatients; registry target 700 UNKNOWN
Double-blind, randomized, multicenter parallel registry trial; no results located in current registry snapshot PLACEBO
UNKNOWN
NO RESULTS LOCATED; registry last updated 2022-04-05; current status UNKNOWN Composite clinical worsening: dyspnea/fatigue worsening, oxygen, hospitalization or death within 12-14 days Safety: UNKNOWN
YES Trial registry explicitly identifies DrinkHRW as the source of study magnesium tablets
WITHIN_ARM_BENEFITS_REPORTED; no significant between-schedule differences; not evidence of HRW versus no-H2 efficacy Fatigue Severity Scale; secondary SF-36 physical function and DASS; salivary uric acid Safety: UNKNOWN
YES University source lists commercial funder for the CFS hydrogen-water project; investigator review identifies magnesium-based tablet preparation
Dementia patients; sample size not reported in accessible abstract UNKNOWN
Controlled human intervention report; dementia patients received electrolysis-produced hydrogen water and purified water for six months; allocation details/sample size not in accessible abstract PURIFIED_WATER
UNKNOWN
Multiple biomarker benefits reported; no prespecified single primary outcome identified Serum AD biomarkers, inflammatory/Th1-Th2 cytokines, ROS and nitric oxide Safety: UNKNOWN
UNKNOWN Funding/COI details not in accessible abstract
Limitations – what this research still cannot tell me
This is a structured review, not an exhaustive systematic review or meta-analysis. The confirmed cohort contains 54 families; 197/204 candidates remain insufficient or unresolved. Further adjudication may add trials and alter descriptive counts.
Only 10/54 families have numeric concentration entries, with mixed measurement bases; point-of-use concentration is often unverified. Primary-outcome identification, prespecification, effect sizes, safety and disclosures are incomplete for some studies. The brand sample is purposive. Citation counts do not measure evidence quality, and “not identified” does not mean “does not exist.”
Studies cited in the claims table
[1] Aoki K et al. (2012). Pilot study: Effects of drinking hydrogen-rich water on muscle fatigue caused by acute exercise in elite athletes. PMID 22520831.
[2] Hori A et al. (2020). Two-week continuous supplementation of hydrogen-rich water increases peak oxygen uptake during an incremental cycling exercise test in healthy humans. PMID 33380582.
[3] Valenta M et al. (2022). Acute pre-exercise hydrogen rich water intake does not improve running performance at maximal aerobic speed in trained track and field runners. PMID 36538554.
[4] Ogannisyan M et al. (2025). Hydrogen-Rich Water Decreases Muscle Damage and Improves Power Endurance in Elite Athletes. PMID 40376695.
[5] Sim M et al. (2020). Hydrogen-rich water reduces inflammatory responses and prevents apoptosis of peripheral blood cells in healthy adults. PMID 32699287.
[6] LeBaron TW et al. (2020). The Effects of 24-Week, High-Concentration Hydrogen-Rich Water on Body Composition, Blood Lipid Profiles and Inflammation Biomarkers in Men and Women with Metabolic Syndrome. PMID 32273740.
[7] Does Hydrogen-Rich Water Reduce Oxidative Stress in Patients with Chronic Mountain Sickness? Randomized, blinded, controlled trial. PMID 41371770.
[8] Kajiyama S et al. (2008). Supplementation of hydrogen-rich water improves lipid and glucose metabolism in patients with type 2 diabetes or impaired glucose tolerance. PMID 19083400.
[9] Hydrogen-Rich Water Ameliorates Metabolic Disorder via Modifying Gut Microbiota in Impaired Fasting Glucose Patients. PMID 37371975.
[10] Song G et al. (2013). Hydrogen-rich water decreases serum LDL-cholesterol levels and improves HDL function in patients with potential metabolic syndrome. PMID 23610159.
[11] Hydrogen Activates ATP-Binding Cassette Transporter A1-Dependent Efflux Ex Vivo and Improves High-Density Lipoprotein Function in Patients With Hypercholesterolemia. PMID 25978109.
[12] Todorovic N et al. (2025). HYDRAPPET: Effects of 8-Week Hydrogen-Rich Water Consumption on Appetite, Body Composition, Sleep Quality, and GLP-1 in Obese Men and Women. PMID 40731927.
[13] Hydrogen-rich water reduces liver fat accumulation and improves liver enzyme profiles in patients with non-alcoholic fatty liver disease. PMID 30982748.
[14] Effects of Molecular Hydrogen Assessed by an Animal Model and a Randomized Clinical Study on Mild Cognitive Impairment. PMID 29110615.
[15] Yoritaka A et al. (2013). Pilot study of H₂ therapy in Parkinson's disease: a randomized double-blind placebo-controlled trial. PMID 23400965.
[16] Yoritaka A et al. (2018). Randomized, double-blind, multicenter trial of hydrogen water for Parkinson's disease. PMID 30207619.
[17] Peripheral endothelial function can be improved by daily consumption of water containing over 7 ppm of dissolved hydrogen: A randomized controlled trial. PMID 32470022.
[18] Tan Y et al. (2024). The Effect of 14-Day Consumption of Hydrogen-Rich Water Alleviates Fatigue but Does Not Ameliorate Dyspnea in Long-COVID Patients. PMID 38794767.
[19] Aker MN et al. (2024). The effect of hydrogen-rich water consumption on premenstrual symptoms and quality of life: a randomized controlled trial. PMID 38532373.
[20] Kang KM et al. (2011). Effects of drinking hydrogen-rich water on the quality of life of patients treated with radiotherapy for liver tumors. PMID 22146004.
[21] Gaboreau Y et al. (2024). Molecular Hydrogen for Outpatients with COVID-19 (Hydro-COVID): A Phase 3 Randomised, Triple-Blinded, Placebo-Controlled, Multicentre Trial. PMID 39124575.
[22] Zanini D et al. (2021). The effects of 6-month hydrogen-rich water intake on molecular and phenotypic biomarkers of aging in adults aged 70 years and over: a randomized controlled pilot trial. PMID 34601077.
[23] Todorovic N et al. (2025). Effects of 12-Week Hydrogen-Rich Water Consumption on Body Composition and Metabolic Biomarkers in Adults with Metabolic Syndrome Stratified by Breath Hydrogen Levels: A Randomized Controlled Trial. DOI 10.37290/ctnr.v23i2.29; registry NCT06968715.
How we test products
Every product we review is tested using the same structured process, with dozens of measurements and checks covering areas such as thermal performance, lid usability, cleaning, durability, portability, and everyday use. Those results feed into the WBA Score, which helps us compare bottles using consistent criteria instead of relying only on manufacturer claims.
Jeremiah Kowalski founded Water Bottle Advisor to make drinkware reviews more useful and less dependent on manufacturer claims. He has personally tested 50+ products, putting them through measuring, cleaning, leak testing, durability testing, and everyday use before sharing his conclusions. It also means occasionally explaining why yet another large box of bottles has arrived at his door.
Leave a Reply