Tip · Breathing

Wim Hof Breathing

Q: Why must this not be done in water?

Because the drive to breathe runs via CO2 and not via oxygen. Hyperventilation lowers CO2 and thereby delays the signal to take a breath while oxygen keeps falling, so that loss of consciousness comes without warning. The American Red Cross has issued a recommendation of the highest level of obligation on this. The CDC has documented four deaths among advanced to excellent swimmers aged 17 to 22 who had deliberately hyperventilated beforehand.

Q: How far does oxygen really fall?

In the only study with continuous measurement, fingertip pulse oxygen saturation fell to a mean of 60 percent at the end of the breath holds, and that in healthy runners sitting on land. At the same time, CO2 had been washed out so far by the preceding hyperventilation that the urge to breathe set in with a strong delay. It is exactly this combination that makes the exercise fatal in water.

Q: Is it true that the immune effect rests on only twelve people?

The original 2014 study had twelve people in the intervention group and twelve in the control group. But it is no longer the only data source: in 2022 the same researchers repeated the investigation with 48 participants and showed that the effect is attributable to the breathing exercise and also occurs when an independent trainer guides it. All participants in both studies were young healthy men.

Q: Does the method do anything for blood pressure or stress resilience?

According to the available data, no. A randomized trial with 42 participants over 15 days found no group differences whatsoever in blood pressure, vascular stiffness, heart rate, HRV, perceived stress, affect or vitality. In the largest blinded trial with 200 participants, stress improved just as much in the breathing group as in the placebo group.

Q: Does the breathing help with sports?

Not acutely. A session before a sprint test changed neither peak power nor mean power nor the fatigue index, despite dramatic blood gas effects. The systematic review of all studies describes the performance data as inconsistent overall.

Q: Is the cold or the breathing the active part?

For the anti-inflammatory effect, this has been studied directly. Cold training alone did not relevantly change the inflammatory response to endotoxin, the breathing exercise alone did, and cold further amplified the effect of the breathing exercise. The combination had the strongest effect.

Hyperventilation plus cold. The immune effect comes from a study with twelve participants. Never practice in or near water.

Wim Hof breathing is cyclic hyperventilation followed by a breath hold. It produces one of the strongest bodily reactions that can be triggered at will — and it is the only technique in this group where the wrong circumstances end fatally.

In short

The breathing exercise releases adrenaline and thereby dampens the inflammatory response to an endotoxin dose. The finding is genuine and has since been reproduced: first 12 participants against 12 controls, then a follow-up study with 48 — all young, male, healthy. On blood pressure, HRV, perceived stress and sprint performance, by contrast, nothing could be shown, and against a cleanly designed placebo with 200 participants the psychological benefit disappeared. The safety question is decisive: the combination of hyperventilation and breath hold pushed oxygen saturation down to 60 ± 12%, while the urge to breathe fails to appear. Therefore never in water, at the poolside, in the bathtub or before swimming — that is exactly where people have died.

What is behind it

The exercise consists of two opposing phases. In the hyperventilation phase, CO2 is washed out: end-tidal CO2 partial pressure fell to 19 ± 3 mmHg, and arterial pH rose by an estimated +0.171 — a pronounced respiratory alkalosis. In the subsequent breath hold, oxygen falls steeply: fingertip pulse saturation reached a mean of 60 ± 12% at the end of the holds, in healthy amateur runners, seated, in the laboratory, on land.

How it works

Alkalosis and intermittent hypoxia lead to a strong release of adrenaline. Adrenaline increases the release of the anti-inflammatory cytokine IL-10 and suppresses TNF-α, IL-6 and IL-8. The correlation between adrenaline levels and the rise in IL-10 is explicitly described in the original paper.

Which component carries the effect has been studied: cold training alone did not relevantly change the inflammatory response (p = 0.77), the breathing exercise alone produced the effect (p = 0.002), and cold amplified it further (p = 0.02).

Never in or near water

The drive to breathe is controlled not by oxygen deficiency but by the rise in CO2; the breathing reflex kicks in at an arterial CO2 partial pressure of about 45 to 60 mmHg. Hyperventilating beforehand artificially lowers CO2 and delays exactly the signal that would otherwise force you to take a breath. Oxygen nonetheless keeps falling during the breath hold, very quickly at the end of the binding curve. Loss of consciousness occurs when arterial oxygen partial pressure drops below about 25 to 30 mmHg — without any urge to breathe being felt beforehand.

This is not a theoretical danger. For New York State, the CDC recorded 16 drowning events between 1988 and 2011 associated with dangerous underwater breath-holding behaviors: 4 fatal, 12 nonfatal. All four who died were between 17 and 22 years old, were considered advanced to excellent swimmers and had hyperventilated deliberately. In an Australian analysis of 317 deaths while snorkeling and breath-hold diving, at least 137 were primary drownings, 34 of them following apnea-related hypoxia.

The Scientific Advisory Council of the American Red Cross has published a recommendation of the highest strength category on this: voluntary hyperventilation before underwater swimming or underwater breath holding is a dangerous activity that lifeguards and swim instructors should prohibit for all persons.

What is well supported

The immune finding holds up. In the original study, healthy volunteers were randomized to an intervention group (n = 12) and an untrained control group (n = 12); after ten days of training, all received 2 ng/kg E. coli endotoxin intravenously. While the trained participants applied the technique, alkalosis and hypoxia developed with markedly elevated plasma adrenaline: IL-10 rose faster and higher, TNF-α, IL-6 and IL-8 were lower, and flu-like symptoms were milder. The same group repeated this in standardized form in 2022 — 40 healthy men for the adrenaline endpoint, 48 for the cytokine endpoint — and the rise in adrenaline depended neither on the duration of training (p = 0.71) nor on the trainer (p = 0.46).

What the studies show

The endotoxin study (Kox et al. 2014)

12 healthy volunteers per group. The authors’ interpretation: voluntary activation of the sympathetic nervous system releases adrenaline and suppresses the innate immune response.

Blood gases and sprint performance (Citherlet et al. 2021)

Randomized crossover with 15 amateur runners: one session of Wim Hof breathing versus pure hyperventilation and spontaneous breathing. End-tidal CO2 fell to 19 ± 3 mmHg and 17 ± 3 mmHg respectively, corresponding to estimated pH increases of +0.171 and +0.181. Despite these massive effects, the session improved neither peak power nor mean power nor the fatigue index in the Repeated Ability Sprint Test.

The placebo test (Fincham et al. 2024)

The largest blinded randomized trial to date on hyperventilation breathing techniques with retention: 200 young healthy adults, three weeks of 20 minutes daily of fast breathing with long breath holds or a placebo comparison at 15 breaths per minute. Both conditions were masked, and only about 40% guessed their group. The primary endpoint was missed: no group-by-time interaction for stress (p = 0.16), but a clear time effect (p < 0.01) in both groups.

Where the data stop

The immune studies are small and one-sided in composition: 12 people per group in the original study, exclusively healthy male volunteers in both sub-studies of the follow-up paper. There are no data on women, older people or people with pre-existing conditions. The endpoint is also a surrogate: the cytokine response to an artificial endotoxin dose in the laboratory, not the frequency of infections or a hard clinical endpoint.

Everything that goes beyond the inflammatory response has not so far withstood testing. Over 15 days, the full method had no measurable effect in 42 participants on blood pressure, vascular stiffness, heart rate, HRV or any of the psychological parameters assessed (all p ≥ 0.186). Against a placebo with 200 participants, stress improved equally in both groups. For athletic performance the finding is negative. And participants fundamentally cannot be blinded to a Wim Hof intervention.

How to do it

In the studies, it ran over ten days of training in three components: meditation, breathing techniques with cyclic hyperventilation followed by breath holding, and cold exposure. A single cycle in the sprint study looked like this: hyperventilation phase, then a breath hold at low lung volume for as long as possible, followed by inhaling to total lung capacity and holding for 15 seconds. It is exactly this breath hold that pushes saturation down to 60 ± 12% — it should be clearly named and not put in brackets, because it is the dangerous part.

Safety

For this tip, this section is not the appendix but the core. The rule is: never in water, never at the poolside, never in the bathtub, never before swimming or diving. Not while driving, cycling, on ladders, on stairs or in any place where a fall would be dangerous. Only lying down or sitting with your head secured; in the studies, practice was guided and supervised. Loss of consciousness and convulsions are not a rare side effect here but the foreseeable consequence if you go far enough — and because the CO2 has been washed out beforehand, it comes without warning. Dizziness and tingling in the hands, feet and around the mouth are part of the alkalosis; in a randomized trial, 10% of participants reported negative experiences, and the arm with cyclic hyperventilation and retention was the largest there, with n = 32. For pregnant women, children and adolescents, and for people with epilepsy, coronary heart disease, cardiac arrhythmias, untreated high blood pressure, aneurysms, pulmonary hypertension, glaucoma or panic disorder there are no safety data; the existing studies included exclusively young, healthy, non-smoking participants with an unremarkable ECG. Missing data are not an all-clear.

BK-Score Thin human evidence

Human evidence		4
Mechanism		6
Safety data		4
Hype gap		3
Track record of use		7

The endotoxin study from Nijmegen with twelve trained participants was reproduced by the same group with 48 healthy men – beyond that there is hardly anything, and all participants were young, male and healthy. The safety data score is the lowest value on this list: hyperventilation followed by breath holding has repeatedly led to deaths in water, and on land saturation falls to around 60%.

The score rates the state of knowledge, not the effect. “Safety data 9” means well studied – not harmless.
Subjective assessment by Biohacking Kompakt based on published scoring rules – not a scientific rating and not a medical recommendation. Rules and all ratings (German)

Frequently asked questions about Wim Hof Breathing

Why must this not be done in water?