Two independent university reports have been produced on the H2Plus water source that ATPEAK is built on. Both are published here in their entirety — the methods, every outcome that was measured, the exact p-values, the results that came back null, and the limitations the researchers themselves wrote down.
Both studies were conducted in horses. No human efficacy trial has been run. We think you should know that before you read anything else on this page.
Performance-beverage marketing has a habit of quoting one impressive number and burying everything around it. We would rather hand you the reports and let you check our work — including the parts that don't help us.
If you are a researcher, a retailer's compliance team, a journalist, or just someone who wanted to know whether the number on the bottle means anything, this page is written for you.
Two studies exist. Here is what each one actually found.
Oklahoma State (ex vivo). Researchers took muscle biopsies from four Thoroughbred geldings, isolated the mitochondria, and put them in a dish — some in H2Plus-produced water, some in ultrafiltered laboratory water. The mitochondria bathed in H2Plus water produced ATP 25% faster (p = 0.005). That is a real and statistically strong finding. It is also a finding about isolated horse mitochondria in a dish, not about a person drinking a bottle of water.
Washington State (in vivo). Eight fit Thoroughbreds drank either H2Plus water or Pullman tap water for roughly three weeks, then ran treadmill tests, then swapped. Seven of the eight ran longer to fatigue on H2Plus water. Across all eight horses that difference was not statistically significant (p = 0.07). It reached significance (p = 0.02) only after the researcher removed one horse's data — the one horse that ran longer on tap water, on a day the report notes was cooler. Maximal oxygen uptake also fell short of significance (p = 0.06).
So: a strong cellular result in a dish, and a directionally encouraging but statistically inconclusive result in living animals. Nothing in humans.
Earlier versions of our marketing quoted the Washington State stamina result as "p = 0.02" without mentioning that the figure comes from a seven-horse analysis after one animal was excluded. The all-horses number is p = 0.07. We also quoted a "13% increase in maximal phosphorylating respiration" alongside a significant p-value in a way that implied the 13% figure was itself significant. It was not — that comparison returned p = 0.0653.
Both are corrected here, and this page is now the authoritative version of what the studies found.
Report on Ex Vivo Testing of H2PLUS Technologies Proprietary Water on Skeletal Muscle Mitochondrial Function
| Measure | H2Plus | Control | Change | p-value | Verdict |
|---|---|---|---|---|---|
| Maximal ATP synthesis rate pmol ATP/(s·ml) |
419.65 | 335.43 | +25% | 0.005 | Significant |
| Complex II–supported respiration pmol O₂/(s·ml) |
85.67 | 70.66 | +21% | 0.0285 | Significant |
| Mitochondrial efficiency calculated, % |
96.50% | 96.05% | +0.45 pts | 0.0416 | Significant |
| Maximal phosphorylating respiration pmol O₂/(s·ml) |
134.4 | 119.1 | +13% | 0.0653 | Not significant |
| Complex I–supported respiration pmol O₂/(s·ml) |
80.64 | 76.07 | +6% | 0.4501 | Not significant |
| Oxidative phosphorylation efficiency (P/O ratio) pmol ATP/pmol O₂ |
7.25 | 7.05 | +3% | 0.346 | Not significant |
| Leak respiration pmol O₂/(s·ml) |
4.444 | 4.505 | −1% | 0.6667 | Not significant |
| Leak respiration flux control ratio % |
6.467% | 6.679% | −0.21 pts | 0.6909 | Not significant |
| H₂O₂ production — Complex II respiration nmol H₂O₂/pmol O₂ |
0.00106 | 0.00202 | −48% | 0.089 | Trend only |
| H₂O₂ production — leak state nmol H₂O₂/pmol O₂ |
0.02078 | 0.02709 | −23% | 0.200 | Not significant |
| H₂O₂ production — Complex I respiration nmol H₂O₂/pmol O₂ |
0.00097 | 0.00078 | +24% | 0.320 | Not significant |
| H₂O₂ production — max phosphorylating nmol H₂O₂/pmol O₂ |
0.00043 | 0.00050 | −14% | 0.345 | Not significant |
Tissue handling. Biopsies were transferred immediately into ice-cold BIOPS solution (2.77 mM CaK₂-EGTA, 7.23 mM K₂-EGTA, 20 mM imidazole, 20 mM taurine, 50 mM K-MES, 0.5 mM dithiothreitol, 6.56 mM MgCl₂, 5.77 mM ATP, 15 mM phosphocreatine, pH 7.1, prepared with ultrapurified laboratory water) and transported to the laboratory. Mitochondria were isolated with a commercial kit (MITOISO1, Sigma-Aldrich), yielding roughly 80 µl of mitochondrial suspension per 100 mg of fresh tissue. Samples were held at 0–4 °C until 15 µl was added to each respirometry chamber, and all analysis was completed within 12 hours of biopsy.
Respiration media. 0.5 mM EGTA, 1 mM MgCl₂, 60 mM K-lactobionate, 20 mM taurine, 10 mM KH₂PO₄, 20 mM HEPES, 110 mM sucrose, 1 g/l essentially fatty-acid-free BSA, pH 7.1 — prepared with either H2Plus water or the control solvent.
SUIT protocol. After injection of the mitochondrial suspension and signal stabilisation: pyruvate (5 mM), glutamate (10 mM) and malate (0.5 mM) were titrated in to produce NADH and stimulate non-phosphorylating leak respiration through Complex I (LN). ADP was added to stimulate phosphorylating respiration through Complex I (PN). Succinate (10 mM) produced phosphorylating respiration through Complex I and II combined (PN+S). Rotenone (0.5 µM) blocked Complex I, leaving Complex II capacity alone (PS). All oxygen flux measurements were corrected for residual oxygen consumption.
Derived measures. Flux control ratios were calculated for leak respiration, NADH-supported and succinate-supported respiration, and oxidative phosphorylation efficiency (1 − LN/PN+S). Maximal ATP synthesis was derived from the change in Magnesium Green fluorescence during the PN+S phase, using a per-chamber MgCl₂ calibration and condition-specific Kd values for ADP and ATP binding with Mg²⁺, following Chinopoulos et al. 2014. Hydrogen peroxide rates used Amplex UltraRed calibrations obtained periodically in each chamber, following Krumschnabel et al. 2015.
Analysis. Results were calculated per chamber, then duplicates averaged by subject within each condition. Oxygen consumption used repeated-measures two-way ANOVA (Prism 9.1.0) with horse as the repeating variable and solvent and fluorophore as independent variables; where the overall ANOVA returned p < 0.1, post-hoc pairwise comparisons used Fisher's Least Squares Difference. ATP synthesis and hydrogen peroxide rates were compared with paired Student's t-tests.
Mechanism. The report attributes the primary effect to increased Complex II activity, with a smaller effect somewhere in the common downstream pathway (ubiquinone, Complex III, cytochrome c, or Complex IV). Because respiration through Complex I involves more parallel reactions, the authors state they cannot completely rule out an effect there that went undetected.
On efficiency. Leak respiration did not change, so the authors reason the efficiency gain likely reflects improved function of the elements that move protons across the inner membrane, rather than reduced proton leakage. The increase in ATP synthesis rate is described as the result of greater proton motive force driving ATP synthase, not a change in ATP synthase itself.
On reactive oxygen species. The report describes the impact on ROS generation as limited. Only the Complex II comparison approached anything (p = 0.089), and the authors call it a trend, not a finding.
The limitation in the authors' own words. "The primary limitation of this study is the low statistical power that is inherent in a study with a small number of subjects. With only 4 subjects, there is a chance that relevant biological effects of the H2PLUS water were not detected statistically." They add the converse — that detecting significant changes at n = 4 at all "highlights the magnitude and relative uniformity of these changes."
The 25% ATP result is the strongest number in the entire research record, and it is not a fluke of a marginal p-value — 0.005 is solid. But it describes isolated mitochondria sitting in the water, in a dish, from four horses. It is a mechanistic finding. It is not evidence that a person who drinks ATPEAK makes 25% more ATP, and we should not have implied otherwise.
Ergogenic Evaluation of Exercise Performance in Response to Ingestion of H2Plus Technologies Water in Horses · April 2022
| Measure | H2Plus | Tap water | n | p-value | Verdict |
|---|---|---|---|---|---|
| Run time to fatigue — all horses | — | — | 8 | 0.07 | Not significant |
| Run time to fatigue — after excluding one horse seconds, mean ± SD |
1047 ± 232 | 908 ± 182 | 7 | 0.02 | Significant* |
| Maximal oxygen consumption (V̇O₂max) | — | — | 7 | 0.06 | Not significant |
| Speed at which V̇O₂max was reached m/s, mean ± SD |
9.8 ± 1.1 | 9.9 ± 0.7 | 7 | 0.79 | No difference |
| Relative intensity of the fatigue run % of V̇O₂max, mean ± SD |
77.3 ± 9.9% | 77.8 ± 13.2% | — | 0.99 | No difference |
| Post-exercise blood lactate mmol/l, mean ± SD |
7.3 ± 3.6 | 5.9 ± 3.5 | — | 0.09 | Not significant |
The p = 0.02 stamina result is the number our marketing has quoted. It comes from a seven-horse analysis. The eighth horse — the only one that ran longer on tap water — was removed because, in the report's words, "there was a cooler ambient temperature that day." With all eight horses included, the comparison returns p = 0.07, which does not meet the study's own significance threshold.
Removing a data point after seeing the result is a post-hoc exclusion. There may well be a defensible physiological reason for it, and the reasoning is stated openly in the report rather than hidden. But the honest way to present this finding is with both numbers, and that is what we do here.
Separately, the report describes the p = 0.06 V̇O₂max result as meaning "a 94% certainty that drinking H2Plus Technologies increased V̇O₂max." That is a common way of speaking about p-values, but it is not what a p-value means — see the note below.
Two unfit horses drank H2Plus water for 21 days, with complete blood counts and equine plasma ion and biochemistry panels drawn at day 0, day 10 and day 21, and daily water intake monitored.
The report found no sign of any haematological or serum biochemical abnormality associated with 21 days of consumption. Water consumption was normal at 10–16 gallons per day. Several values in the panel sat slightly outside the stated reference ranges at various timepoints in both animals — including packed cell volume, haemoglobin, creatine kinase and albumin — with no consistent direction attributable to the water.
This is a two-animal, three-week tolerability check. It is reassuring as far as it goes, and it goes about that far.
"The horses' health was not adversely affected by drinking the H2Plus Technologies water… The horses were fit when they began the study and there was no apparent improvement or decrease in their aerobic capacity over the course of the study. Despite this, their V̇O₂max was increased (94% probability) and their endurance was enhanced as reflected by the marked increase in the run time to fatigue in 7 of the 8 horses."
The report also notes that four horses were able to run at 12 m/s after drinking H2Plus water, while only one control horse reached that speed and held it for just 20 seconds. It proposes reduced reactive oxygen species in muscle as a possible mechanism, citing the muscle-biopsy component of the same investigation — though as Study 01 shows, the ROS findings there did not reach significance.
Seven of eight horses running longer is a real signal and worth taking seriously. It is also, by the study's own threshold, an inconclusive result at n = 8. The honest description is "directionally encouraging, not established" — which is a perfectly respectable thing for an early-stage study to be. It is not a proven performance benefit, in horses or in anyone else.
Our marketing has referred to a third study at the University of South Carolina, described as a safety and tolerability assessment in physically active adults, and attributed to Susan Yeargin, PhD, ATC.
We have not published that report on this page, and until we do you should treat our summary of it as unverified. When we have the full report and permission to post it, it will appear here alongside the other two, with the same treatment.
Two things worth stating plainly in the meantime. First, that study addressed whether the water is safe to drink — it was not a test of whether it improves performance, and it should never be cited as one. Second, "USC" in a fitness context reads as the University of Southern California; the reference is to the University of South Carolina, and we have corrected that ambiguity wherever we found it.
The US Federal Trade Commission's guidance on health-related advertising holds that in vitro work "is of limited value to predict benefits for humans," and that animal studies, without confirmation by human randomized controlled trials, "aren't sufficient to substantiate health-related claims." Both of our studies fall into those categories. We hold ourselves to that standard here.
A randomized, controlled, adequately powered human trial measuring a real performance endpoint. That is the study that would let us make a claim about people, and it is the study that does not exist yet. If and when it is run, the results will be posted on this page whichever way they come out.
A p-value is the probability of seeing a difference at least this large if the treatment actually did nothing. A small p-value means the result would be surprising under that assumption. It is not the probability that the treatment works, and 1 minus the p-value is not "the certainty that it works" — which is why we flagged the "94% certainty" phrasing in the Washington State report rather than repeating it.
By convention, p ≤ 0.05 is called statistically significant. That threshold is arbitrary but useful: it is the line both of these studies set for themselves, so it is the line we hold their results to. A result at p = 0.07 is not "nearly significant" in any meaningful sense — it is a result that did not clear the bar the researchers chose in advance.
Sample size matters enormously here. Four horses and eight horses are small studies. Small studies miss real effects, and they also produce unstable estimates of the effects they do find. Both of these reports say so themselves. That cuts in both directions, and we have tried not to claim only the direction that flatters us.
Complete, unedited, as delivered by the researchers.
Report on Ex Vivo Testing of H2PLUS Technologies Proprietary Water on Skeletal Muscle Mitochondrial Function. Ex vivo respirometry, four Thoroughbred geldings.
Download PDFErgogenic Evaluation of Exercise Performance in Response to Ingestion of H2Plus Technologies Water in Horses. Warwick Bayly, April 2022. Randomized crossover, eight Thoroughbreds.
Download PDFQuestions about methodology, or want to discuss running a human trial? Write to research@atpeak.vip. Researchers named in these reports conducted the work; their inclusion here is a citation of published findings and is not an endorsement of ATPEAK by them or by their institutions.
The research is early, it's in horses, and we've told you exactly where it falls short. If that's the kind of company you want to buy from, the bottle is here.
Shop ATPEAK