Slow, paced breathing has real trial evidence behind it for stress and anxiety, though the trials are mostly small. Some popular claims around it are far ahead of the evidence: “30-second resets”, breath-hold scores, mouth taping.
Evidence: Mixed slow breathing / HRV biofeedback · Mixed cyclic sighing (one trial) · None BOLT cut-offs
Breathing is one of the few automatic body functions you can also control at will. That makes it an easy thing to practise. It doesn’t make every claim about it true, so this post grades them one at a time.
The grades in this post: Strong means good randomised trials agree. Mixed means the trials are few, small, short or unblinded, or disagree. Observational means studies show a link, not proof of cause. Mechanistic means the biology or animal work makes sense but it hasn’t been shown to change outcomes in people. None means there’s no usable evidence in people.
Breathing in speeds the heart slightly; breathing out slows it. Slow breathing, at around six breaths a minute, makes that swing larger, which shows up as higher heart rate variability (HRV). That is the idea behind HRV biofeedback. Whether a bigger swing is what explains the benefits isn’t proven. Over-breathing also lowers carbon dioxide in the blood, which makes haemoglobin hold on to oxygen more tightly (the Bohr effect). That is textbook physiology; whether it explains everyday tiredness or brain fog hasn’t been tested.
Most of the evidence here comes from HRV biofeedback: slow breathing, usually around six breaths a minute, guided by a heart-rate display. Across 24 studies, it was linked to a large drop in self-reported stress and anxiety, but the authors said better-controlled trials are still needed (Goessl 2017). A wider review of 58 randomised trials found a small-to-moderate benefit across many outcomes. Effects were biggest for anxiety, depression and anger, and smallest for PTSD, sleep and quality of life. Effects were also bigger against “do nothing” controls than against active ones (Lehrer 2020). For depressive symptoms, 14 trials gave a moderate effect (Pizzoli 2021). Mixed
Nobody has tested it head-to-head against CBT, so I won’t claim it’s “as good as” CBT. It’s a reasonable thing to try alongside proper care.
The BOLT (Body Oxygen Level Test) is a comfortable breath hold after a normal breath out. It’s popular in the Buteyko and McKeown world. It hasn’t been scientifically validated (Kowalski 2024). In two studies it didn’t track fitness or exercise performance at all (Kowalski 2024; Marko 2025). I don’t know of any evidence behind the usual cut-off scores, so I don’t use them. None
A cyclic sigh is a double breath in through the nose and a long, slow breath out. In a Stanford trial, people practised one of four techniques for five minutes a day for a month. The four were cyclic sighing, box breathing, cyclic hyperventilation, or mindfulness meditation. Breathing practice, especially cyclic sighing, improved mood more and slowed resting breathing more than mindfulness did (Balban 2023). Mixed One randomised trial.
That’s a promising single trial of a daily practice. It doesn’t show that one or two sighs “reset” stress in 30 seconds. The idea that the double breath re-opens collapsed air sacs is a plausible explanation, not something the trial measured.
The Wim Hof method combines fast, deep breathing and breath holds with cold exposure and meditation. In one trial, 12 healthy men trained for 10 days and were then given a bacterial toxin that triggers a short inflammatory response. They had higher anti-inflammatory and lower inflammatory markers, and fewer flu-like symptoms, than 12 untrained men (Kox 2014). It’s one small trial in a lab model, and it can’t tell you whether the breathing, the cold or the meditation did the work. Mixed
The breathing phase deliberately drives your carbon dioxide down and your oxygen low. Don’t do it if you have epilepsy, heart problems or a history of fainting, and never in or near water.
Mouth taping at night is all over social media. A 2025 review found only 10 small studies. Two showed improvements in sleep apnoea measures, others showed no difference, and the authors flagged a real risk of harm if your nose is blocked (Rhee 2025). If you snore or mouth-breathe at night, get your nose and your sleep checked first. Mixed With a safety concern.
How you feel, how you sleep, and your resting breathing rate. Those are what the trials measured. Wearable HRV can be interesting to watch, but it hasn’t been validated as a measure of whether breathwork is “working”. I don’t use stress hormone tests or blood inflammation markers to judge breathing practice. Nobody has tested them for that.
Slow breathing is free, low-risk and has decent small-trial support for stress. That’s enough reason to try it. It’s not a reason to believe everything sold alongside it.
| Claim | Grade | Main source |
|---|---|---|
| Slow breathing / HRV biofeedback for stress and anxiety | Mixed many small trials | Goessl 2017; Lehrer 2020 |
| HRV biofeedback for depressive symptoms | Mixed 14 trials, moderate effect | Pizzoli 2021 |
| Cyclic sighing (5 min a day) for mood | Mixed one randomised trial | Balban 2023 |
| One sigh “resets” stress in 30 seconds | None not tested | — |
| Wim Hof breathing, cold and meditation dampen the immune response | Mixed one small trial, lab model | Kox 2014 |
| BOLT score cut-offs | None not validated | Kowalski 2024; Marko 2025 |
| Mouth taping at night | Mixed 10 small studies; safety concern | Rhee 2025 |
| Breathwork is as good as CBT | None no head-to-head trial | — |
| Hormone or inflammation tests can judge breathwork | None untested | — |
I sell testing, consultations and personal training, so I declare an interest: the article’s grades apply to them too.
Ask the DH Clinical Concierge — built on nearly forty years of working with people. Free, open, no account required.
Ask the Concierge → Book a call with Stephen