Search "is hydrogen water safe" and most of what comes back reassures you that molecular hydrogen (H2) itself is inert and non-toxic. That part is true, and it's not really the interesting question. The more useful question is the one almost nobody answers: if you're running an electrical current through tap water to make hydrogen, what else does that current make?
Tap water isn't pure H2O. It carries trace chlorine, added on purpose for sanitation, along with dissolved minerals. Push electrolysis through it and you can, in principle, generate more than hydrogen — chlorine compounds and ozone can form as byproducts of the same reaction, depending on how the device is built. Both are oxidants, which is a strange thing to have along for the ride in a drink you're taking specifically for its antioxidant effect. Whether this actually happens, and how much, is a testable question. In 2021, two Japanese researchers tested it directly.
What Electrolysis Does Besides Make Hydrogen
Electrolysis splits water into hydrogen and oxygen using two electrodes. At the cathode, water gets reduced into H2 gas — the part you want. At the anode, water gets oxidized into oxygen, but if the water contains chloride ions (which tap water does, from sanitation chemicals), that same anode reaction can also generate chlorine gas or hypochlorous acid. Under the right conditions, oxygen radicals forming at the anode can combine into ozone (O3) as well.
None of this is exotic chemistry — it's the same basic electrochemistry used in water treatment and pool sanitation, just running in miniature inside a bottle. The open question for any given hydrogen water device is how much of it happens, and whether it stays low enough to be a non-issue.
The Study That Actually Measured It
Toshihisa Hatae and Nobuhiko Miwa published the most direct test of this in Medical Gas Research in 2021 (Hatae & Miwa, "Electrolytic hydrogen-generating bottle supplies drinking water with free/combined chlorine and ozone repressed within safety standard under hydrogen-rich conditions," Med Gas Res. 2021;11(2):61–65). They ran electrolysis on tap water using a reference hydrogen-generating bottle and, separately, three other commercially available hydrogen water bottles, then measured free, bound, and total residual chlorine (using two independent colorimetric methods) and dissolved ozone (using three independent detection methods) at multiple time points.
The reference device came out clean. At 10 minutes of electrolysis, it produced 444 µg/L of dissolved hydrogen — and free chlorine actually measured lower after electrolysis than before, dropping from 0.18 mg/L to 0.12 mg/L. At 30 minutes, hydrogen output rose to an average of 479 µg/L, and chlorine still hadn't increased (0.09–0.10 mg/L, against a 0.11 mg/L tap water baseline). Dissolved ozone stayed below the detection limit (<0.05 mg/L) throughout, well under the 0.1 mg/L safety ceiling used in both the US and Japan. The researchers attribute this partly to the electrode's ultra-smooth, platinum-plated surface, which let chlorine gas escape into the air rather than get trapped as micro-bubbles that react with organic matter in the water, and partly to the single-tank design, which had no membrane separating anode and cathode for ozone to accumulate behind.
The three other commercial bottles told a different story. Using the same test, their chlorine levels climbed the longer electrolysis ran — from a roughly 0.1 mg/L baseline up to 0.2–0.5 mg/L at 30 minutes with a full tank, and 0.5–1.0 mg/L at 30 minutes with a half-full tank. Every one of those readings stayed under the World Health Organization's 5 mg/L drinking-water guideline for chlorine, so nothing here crossed a safety line. But the trend line is the finding: flat (or falling) in one device, rising in the other three. (The study didn't report ozone measurements for those three bottles — only for the reference device.)
Why This Comes Down to Engineering, Not the Hydrogen
It's worth separating two questions that get blurred together. Whether H2 gas itself is safe to drink is already well established — Hatae and Miwa describe it in their own introduction as "highly safe... without any side effect," building on the foundational 2007 finding that hydrogen selectively neutralizes the most reactive oxygen species (hydroxyl radicals, peroxynitrite) without disrupting the reactive oxygen species your cells actually need for normal signaling (Ohsawa et al., Nature Medicine, 2007). That's a settled, separate fact from the byproduct question.
The byproduct question is about the device, not the molecule. A different — and more common — engineering approach to the same problem is SPE/PEM membrane electrolysis, which physically separates the anode and cathode into two compartments joined only by an ion-exchange membrane. Only H+ ions cross the membrane to form H2 on the drinking-water side; the oxygen and any byproduct gases generated at the anode vent out a separate channel instead of dissolving into the water you're about to drink. It's a different mechanism than the single-tank, smooth-electrode design tested in the Hatae study, but it's aimed at the same outcome: keeping anode-side byproducts out of the cup.
This is also why membrane and electrode quality show up as actual spec points rather than marketing filler. True Marqet's H2B1, for instance, is built around SPE/PEM membranes specifically described as helping reduce unwanted byproducts while producing rated output up to roughly 2,400 PPB — that's a design choice aimed at exactly the mechanism above, not a claim that the bottle measures or displays your water's chlorine or ozone level for you. If you want to know your own bottle's byproduct output, that takes a chlorine or ozone test kit, the same category of tool the researchers used — not a hydrogen concentration meter, which only reads dissolved H2.
What This Means If You're Already Drinking Hydrogen Water
The honest summary: in the one published test that's measured this directly, byproduct levels stayed under WHO and US/Japan drinking-water safety limits across every device tried, including the ones that trended upward with longer electrolysis. This isn't evidence that hydrogen water is risky. It's evidence that "hydrogen water bottle" isn't one standardized piece of engineering — it's a category, and devices inside that category behave differently under the same test conditions. That's a more useful thing to know than either "it's all fine" or "be scared of it."
A chlorine or ozone smell after electrolysis — something people sometimes describe as a faint "pool" note — is at least chemically consistent with elevated byproducts, since both compounds have low odor detection thresholds. That's not a measurement, just a signal worth noticing rather than ignoring.
Frequently Asked Questions
Does hydrogen water contain chlorine?
Tap water almost always does, since chlorine is added intentionally for sanitation. Whether electrolysis adds more depends on the device: in the 2021 Hatae and Miwa study, a reference bottle showed no increase in chlorine over 30 minutes of electrolysis (levels even dropped slightly in one measurement), while three other commercial hydrogen bottles showed chlorine rising the longer electrolysis ran. Both stayed under the WHO's 5 mg/L drinking-water guideline in that test.
Is ozone dangerous in hydrogen water?
Ozone is a strong oxidant, and the US and Japan both cap dissolved ozone in drinking water at 0.1 mg/L. In the one published test that measured it, ozone stayed undetectable (below 0.05 mg/L) in the device studied — well under that ceiling. It's a legitimate variable to know about, not a documented problem with hydrogen water generally.
Is molecular hydrogen (H2) itself safe to drink?
Yes. This is separate from the byproduct question and is well established on its own — H2 is biologically inert at the concentrations these devices produce, and your gut bacteria already generate some of it naturally. The mechanism (selectively neutralizing the most reactive oxygen species while leaving beneficial ones intact) was first described by Ohsawa and colleagues in Nature Medicine in 2007.
Why does my hydrogen water sometimes smell different?
Chlorine compounds and ozone both have low odor thresholds, so a faint chlorine or "pool" smell is at least consistent with elevated byproducts if you notice it. Smell isn't a measurement, though — it's a cue to test, not a diagnosis on its own.
Do all hydrogen water bottles produce the same byproducts?
No, based on the only published comparison available. Chlorine trends differed meaningfully between the four devices tested in the 2021 study, even though all of them generated meaningful hydrogen output. Electrode surface finish and tank design (single-tank versus membrane-separated) are the variables the researchers point to.
Can I test my own bottle for chlorine or ozone byproducts?
Yes — with the same kind of test the study used: a chlorine test kit (DPD or orthotolidine-based strips, similar to pool test kits) and a dissolved-ozone test kit or meter. That's different equipment than a hydrogen (PPB) meter, which only measures dissolved hydrogen and won't tell you anything about chlorine or ozone.
None of this is an argument that hydrogen water is risky — the safety margins in the one study that's actually measured it were wide, across every device tested. It's more a reminder that "hydrogen water bottle" describes a category of devices with real engineering differences, and byproduct behavior is one of the more testable ways to tell them apart. If a brand can tell you what its electrolysis chamber is built to keep out of the water, not just what PPB number it hits, that's worth noticing.