On this page
- What does the research say about breathwork benefits?
- How do you use this table?
- What do the evidence levels mean?
- Which breathing exercises have research behind them?
- What stands out in the table?
- Which breathwork claims have no good evidence yet?
- Who should skip this or check first
- How often is the table updated?
- Questions people ask
- The main table as CSV
This page puts the breathwork benefits research in one table: what each breathing technique was tested for, who took part, and how strong the result is. The best supported finding is a small to medium drop in self-reported stress and anxiety. Most other findings are early, and nothing here shows that breathing treats or cures a condition. Use it to see what studies actually tested, not as medical advice.
The table describes studies, not advice for you. Slow, gentle breathing carries the least risk. Fast breathing and breath holds carry the most. Each row says who should check with a doctor first. Please read who should not do breathwork before you start.
What does the research say about breathwork benefits?
Breathwork comes with big promises. Calm. Better sleep. Lower blood pressure. A reset for your nervous system.
Behind each promise there is usually one study, and sometimes none. So I gathered it all in one place. Every row comes from a post on this blog, and I opened and read the sources behind each one. If a claim is not in the table, I either found nothing good or have not checked it yet.
One thing to know before you scroll. Most people in these studies were healthy adults, often university students. A row tells you what happened to them, which may not match you.
How do you use this table?
- Find the technique or the result you care about.
- Check who was studied. If you are not like them (older, younger, pregnant, living with a health condition), the row tells you less about you.
- Read the dose (how much they practised), then the main limit. Small, short studies that rely on questionnaires can show a feeling, not a lasting change.
- Read the doctor first column and the linked post. If any of it applies to you, ask your doctor before you try it.
- Treat a row as a reason to try something gentle, not as a promise.
What do the evidence levels mean?
The levels are my plain-words summary, not a formal grade. They look at how many studies exist, how well they were designed, whether people were randomly assigned, and whether the results agree.
- Moderate: several randomised trials or a review that agree, with limits that are not severe. Nothing in this tracker is stronger than "early to moderate".
- Mixed: pooled results point one way, but the trials disagree a lot or were not tightly controlled.
- Early: a small number of small studies, pilots, single sessions, or reviews with a high risk of bias. A signal worth testing, not a conclusion.
- Weak: very low certainty, one poor-quality set of studies, or results that did not show a benefit.
- No evidence found: see the second table. This is not the same as "it does not work". It means I did not find a good test.
Two terms come up in the table. HRV, or heart rate variability, is the small natural change in time between heartbeats. The letter g is a standard effect size: around 0.2 is small, 0.5 is medium.
Which breathing exercises have research behind them?
Here is the full table. Scroll sideways on a phone.
| Technique | Outcome studied | Dose studied | Who was studied and how many | What was found | Evidence level | Main limit | Who should check with a doctor first | Matching post |
|---|---|---|---|---|---|---|---|---|
| Cyclic sighing | Mood after sessions | 5 min a day for 28 days | 108 adults signed up (30 did cyclic sighing), mostly Stanford psychology students | Mood rose more than after mindfulness meditation, and the gain grew with more days of practice | Early | Exploratory trial, registered late, small uneven groups, practice was self-reported | The trial left out people with heart disease, glaucoma, seizures, pregnancy, psychosis, suicidal thoughts, bipolar disorder or substance use disorders. Check with a doctor if any apply | Cyclic sighing and anxiety |
| Cyclic sighing | Anxiety right now | 5 min a day for 28 days | 108 adults signed up, mostly students | Anxiety scores fell about 3.85 points, about the same as mindfulness (3.95). No difference between groups. General (trait) anxiety did not change | Early | Did not beat mindfulness; exploratory trial | The trial left out people with heart disease, glaucoma, seizures, pregnancy, psychosis, suicidal thoughts, bipolar disorder or substance use disorders. Check with a doctor if any apply | Cyclic sighing and anxiety |
| Cyclic sighing | Anxiety 5 minutes later, in real life | 1 minute whenever tense | 47 university students, 43 of them female, average age about 21 | Anxiety was lower after cyclic sighing than after doing nothing. No difference from box breathing | Early | Pilot study, nobody checked that people did it, expectations may play a part, young and mostly female | The pilot left out people with heart, brain, hormone or lung conditions, and pregnancy. Check with a doctor if any apply | Cyclic sighing and anxiety |
| Box breathing | Anxiety right now | 5 min a day for 28 days | 21 adults in the box breathing group (108 signed up for the trial) | Anxiety fell about 3.75 points, similar to mindfulness (3.95). No difference between groups | Early | Small group, exploratory trial; the mood gain did not beat mindfulness | Same trial exclusions as above. Skip the holds, or ask a doctor, if you have a heart or lung condition, a history of fainting or seizures, or are pregnant | Box breathing evidence |
| Box breathing | Anxiety 5 minutes later, in real life | 1 minute whenever tense, 4 seconds per side | 47 university students, mostly female | Anxiety was lower than after doing nothing; fewer attention errors; slower reaction times | Early | Pilot study, no body measurements, young and mostly female | Same pilot exclusions as above. Skip the holds if unsure | Box breathing evidence |
| Slow breathing, about 6 breaths a minute | Heart rate variability (HRV) while breathing | Under 10 breaths a minute, many studies near 6; sessions of 1 to 30 minutes | 15 studies of healthy adults, 9 to 58 people each | HRV generally rose during slow breathing, most consistently near 6 breaths a minute | Early | The review had no randomised trials; breathing at 6 can push up one HRV measure by itself | Heart or lung condition, pregnancy: ask first | Slow breathing and HRV |
| Slow breathing, about 6 breaths a minute | HRV after weeks of practice | 5 min a day for 28 days (mixed techniques) | 108 adults signed up, mostly students | No clear change in HRV or resting heart rate in any group | Early (no change found) | Not a test of 6 breaths a minute specifically | As above | Slow breathing and HRV |
| Slow breathing at your own best rate | HRV ratio, mood and blood pressure, one session | 15 minutes, once | 95 undergraduates, average age 20, 60 percent female | The own-rate group had a higher LF/HF ratio and better positive mood than quiet sitting. Two other HRV measures did not differ | Early | One session, students only; breathing one breath faster did nearly as well on several measures | The study left out heart disease and medicines that affect heart rate or blood pressure. Ask a doctor if that is you | Resonance breathing |
| HRV biofeedback (breathing at a rate guided by a heart sensor) | Stress and anxiety | Varied (abstract only) | 24 studies, 484 participants (abstract only) | Large drop in self-reported stress and anxiety compared with controls (g = 0.83) | Early to mixed | Abstract only; the authors call for better-controlled studies | Heart condition, pregnancy, lung condition: ask first | Resonance breathing |
| HRV biofeedback | Depression symptoms | Varied (abstract only) | 14 randomised trials, 794 adults (abstract only) | Medium drop in depression symptoms (g = 0.38) | Early to mixed | Abstract only; moderate differences between studies | As above; a diagnosed condition needs proper care | Resonance breathing |
| Breathwork, mostly slow-paced | Self-reported stress | Varied; no clear link between practice time and effect | 12 randomised trials, 785 adults (10 slow, 2 fast-paced) | Small to medium drop in stress compared with comparison groups (g = 0.35) | Early to moderate | None of the trials had a low risk of bias; questionnaires only; not clear for clinical groups | Diagnosed anxiety or depression, panic attacks, heart or lung condition, pregnancy: ask first | Anxiety, stress and depression |
| Breathwork, mostly slow-paced | Self-reported anxiety | Varied | 20 randomised trials of adults | Small to medium drop (g = 0.32) | Early to moderate | Moderate risk of bias, self-reported | As above | Anxiety, stress and depression |
| Breathwork, mostly slow-paced | Self-reported depression symptoms | Varied | 18 randomised trials of adults | Small to medium drop (g = 0.40) | Early to moderate | Moderate risk of bias, self-reported; too soon to recommend for clinical groups | As above | Anxiety, stress and depression |
| 4-7-8 breathing | Self-rated sleep problems | 4 cycles morning and evening for 6 weeks | 48 adults with tinnitus (23 breathing, 25 comparison) | Insomnia Severity Index fell from about 8.4 to 4.4 in the breathing group; the comparison group stayed about the same | Weak to early | People knew which group they were in, self-rated, groups unequal at the start, tinnitus patients only | The trials left out asthma or breathing difficulty, heart conditions and others. Ask first, especially before breath holds or in pregnancy | 4-7-8 breathing and sleep |
| 4-7-8 breathing | Heart rate and blood pressure, one session | 6 cycles, three times, in one session | 43 healthy adults aged 19 to 25, mostly women | Heart rate and top blood pressure number were lower afterwards in both sleep groups. Sleep was not measured | Early | One session, few men | As above | 4-7-8 breathing and sleep |
| Slow breathing without a device | Blood pressure in people with high blood pressure | Mostly 4 to 6 breaths a minute; from one 5-minute session up to 6 months | 13 randomised trials, 1,097 adults | Top number 7.68 mmHg lower, bottom number 4.02 mmHg lower than comparison groups | Mixed | Results varied a lot between trials; short and long trials pooled together; concerns about bias | Always ask your doctor first if you have high blood pressure or take medicine. Never change medicine because of a breathing exercise | Slow breathing and blood pressure |
| Different breathing exercises (11 types) | Blood pressure and heart rate | One session up to 90 days | 15 randomised trials, 1,097 adults, mostly with high or raised blood pressure | Top number about 7.06 mmHg lower, bottom number about 3.43 mmHg lower | Mixed | Every trial was rated unclear or high risk of bias; cannot say which exercise works best | As above | Slow breathing and blood pressure |
| Device-guided breathing | Office blood pressure, compared with a sham device or music | Short trials | 5 trials (individual data for 3) | No clear difference in office blood pressure | Early (no clear gain) | Short trials; device-guided only | As above | Slow breathing and blood pressure |
| Breathing exercises (mostly yoga, 14 of 22 trials) | Quality of life in asthma | Varied; up to 3 months | 22 randomised trials, 2,880 adults with mild to moderate asthma | Quality of life score 0.42 points higher (4 trials, 974 people); the Cochrane reviewers' certainty was moderate | Early to moderate | People cannot be blinded; safety was not measured; no severe asthma; studied as an add-on to usual care | Everyone with asthma: ask your doctor or nurse, and keep your medicines and action plan | Breathing exercises and asthma |
| Breathing exercises | Hyperventilation symptoms in asthma | Measured at 4 to 6 months | 2 trials, 118 people | Nijmegen score 3.22 points lower; no clear difference beyond 6 months in two other trials | Early | Few trials | As above | Breathing exercises and asthma |
| Breathing exercises | Lung function in asthma | Varied | 5 trials, 618 people for FEV1 as percent of expected | 6.88 percentage points higher; no clear difference in litres up to 3 months | Weak | Very low certainty | As above | Breathing exercises and asthma |
| A 5-minute paced breathing break | HRV after work stress | One 5-minute session at 6 breaths a minute | 75 full-time employees, average age about 32 | Higher HRV in recovery than after a body scan or sitting quietly | Early | One lab session; a heart measure, not how people felt | Heart, lung or circulation condition, fainting, epilepsy, pregnancy: ask first | Breathing breaks at work |
| Slow breathing at 6 a minute, different lengths | HRV (RMSSD) during breathing | 5, 10, 15 or 20 minutes, once | 59 adults aged 18 to 26 | Same rise at every length; it faded when breathing stopped | Early | Lab study, one pace, healthy young adults | As above | How long a class should be |
| Slow belly breathing at school | Anxiety in teenagers | 5 min, 3 times a week, 5 weeks | 43 high school seniors (75 invited), average age about 17 | Doable, but no clear change in anxiety scores; taking part fell from about 73 to 47 percent of sessions | Weak | Pilot study, too small, home practice | Asthma or lung condition, heart condition, epilepsy, fainting: ask first. Severe or lasting distress: see a GP | Teens and exam stress |
| Relaxation methods that include breathing (mixed with other methods) | Stress, anxiety and depression in pregnancy | Varied | 32 studies, 3,979 pregnant women | Perceived stress 4.1 points lower, anxiety 5.04 lower, depressive symptoms 2.3 lower | Early (indirect) | Breathing was mixed with muscle relaxation, music and more; big differences between studies; no safety data | Everyone who is pregnant: ask your midwife or doctor first | Breathwork in pregnancy |
| Relaxation and breathing techniques | Pain in labour | Varied | 15 studies, 1,731 women (Cochrane) | May reduce pain and raise satisfaction with pain relief | Weak | Low to very low certainty; safety rarely reported | Ask your midwife or doctor | Breathwork in pregnancy |
| Mouth taping, patches or chinstraps | Sleep, snoring and sleep apnea | Varied | 10 studies, 233 adults in the results table (the abstract says 213) with mouth breathing, snoring or sleep apnea | Weak, inconsistent benefits; only 2 studies in mild sleep apnea showed fewer breathing pauses | Weak | All 10 studies were rated poor quality; safety concerns in 4 studies | Talk to a doctor first. Never with a blocked nose, allergies, sinus problems, sleep apnea or a heart condition | Nose vs mouth breathing |
| Nasal breathing during exercise | Exercise comfort and efficiency | Varied | Review of 30 articles | Workable at moderate effort; some efficiency benefits | Early | Narrative review, one search engine | Breathe the way that feels easiest if you feel short of breath | Nose vs mouth breathing |
| Nasal versus mouth breathing | Breathing muscles and brain function | Varied | Review of 11 studies, 313 people, mostly healthy | Different brain areas and uneven muscle effects | Early | Mixed designs, healthy participants | As above | Nose vs mouth breathing |
How strong is the evidence overall? Early. The best supported rows are small to medium drops in self-reported stress, anxiety and depression across pooled trials, and quality of life in mild to moderate asthma as an add-on to usual care. Most other rows rest on a handful of small studies.
What stands out in the table?
Short-term feelings have the most support. The pooled analysis of 12 trials and 785 adults found a small to medium drop in self-reported stress (Fincham et al. 2023). In the one head-to-head trial, anxiety fell about as much with breathing as with mindfulness meditation (Balban et al. 2023). That trial signed up 108 adults (140 agreed to take part, and 108 were enrolled and randomised); the figure used here is the enrolled one.
Body measures are less clear. HRV rises while people breathe slowly, but it did not change over four weeks in the one trial that measured it. Blood pressure results look encouraging in pooled trials, but the trials disagree a lot, and a stricter comparison found no clear difference. A 2026 analysis pooled 13 trials with 1,097 adults (Cheng et al. 2026). A 2024 analysis pooled 15 trials, also with 1,097 adults (Garg et al. 2024). Both papers say 1,097 in their results. It is a coincidence.
Sleep claims are thin. The trials of 4-7-8 breathing are small, people knew which group they were in, and the results come from questionnaires.
Safety evidence is clearer than benefit evidence. The best-documented risk is blackout when fast breathing is followed by a breath hold, especially in or near water. See who should not do breathwork.
Which breathwork claims have no good evidence yet?
These are popular claims where I found no good test, or found that the evidence does not support the claim as stated. "No evidence found" does not mean it is false. It means nobody has shown it yet.
| Popular claim | What we found | Read more |
|---|---|---|
| 4-7-8 breathing helps you fall asleep faster on the night | No trial measured how fast people fell asleep. Trials looked at sleep scores after weeks, or heart rate after one session | Details |
| The 7-second hold in 4-7-8 matters | No study compared 4-7-8 with the same long exhale and no hold | Details |
| A separate human trial of the "physiological sigh" | None found under that name; the evidence is on cyclic sighing | Details |
| Cyclic sighing is the best exercise for anxiety | In the main trial anxiety fell about the same as with mindfulness | Details |
| Box breathing "resets your nervous system" | Nothing found to back this wording; the authors of the vagus nerve idea describe it as proposed, not proven | Details |
| Breathwork stimulates your vagus nerve | A theory paper whose authors call the mechanism proposed, not proven | Details |
| Breathwork permanently raises your HRV | Rises seen during slow breathing; no change over 28 days in the one trial that measured it | Details |
| Buteyko or special breathing cures asthma | The Cochrane review had one Buteyko trial and one Buteyko with yogic breathing; it did not test a cure | Details |
| Breathwork detoxes the body | Described as a practitioner claim; no study found | Details |
| Taping your mouth shut improves sleep in healthy people | No good trials; studies were of people with snoring or sleep apnea, all rated poor quality | Details |
| A 2-minute breathing break works for teachers and desk workers | Nobody has properly tested it; the closest evidence is one 5-minute lab session | Details |
| Breathing exercises lower exam stress in teenagers | One small pilot found no change in anxiety; most breathwork trials are in adults | Details |
| Fast breathing or breath holds are safe in pregnancy | No trials found; researchers commonly excluded pregnancy | Details |
| There is a best session length or a best number of minutes a day | Reviews and a direct lab test found no extra gain from longer or more practice | Details |
| Breathing exercises replace asthma or blood pressure medicine | Studied only as an add-on, never as a replacement | Details |
Who should skip this or check first
- The table is general information about studies, not advice for you. If you have a heart or lung condition, high blood pressure, asthma, epilepsy, a serious mental health condition, or you are pregnant, ask your doctor before any breathing practice. Many of these trials left those people out, so we know less about them.
- Do not do fast breathing or breath holds in or near water, in a bath, while driving or while standing.
- Never change or stop medicine because of anything in this table.
- If stress, anxiety or low mood is constant or affecting your sleep, work or relationships, speak with a doctor or mental health professional.
How often is the table updated?
We re-check the table when new studies come out and at least every quarter. A new row is added only when I have opened and read the study, and I record whether I read the full text or only the abstract. If two sources disagree on a number, I check the paper and use the one in its results tables. I change an evidence level only when a new study is strong enough to move it, and I note the change below. A Stanford trial registered as NCT06526585 (cyclic sighing and box breathing, 80 planned adults) is due to finish in November 2026 and has no results yet.
Changelog
- 24 October 2026: First edition. 29 rows and 15 entries in the "not enough evidence" table, built from the research and safety posts on this blog. Participant counts checked against the papers: Balban 2023 (108 enrolled, 140 consented), Cheng 2026 and Garg 2024 (1,097 each), Rhee 2025 (233 in the results table; the abstract says 213).
Questions people ask
What are the benefits of breathwork, according to research?
The best supported finding is a small to medium drop in self-reported stress, anxiety and depression in pooled trials of adults. Quality of life in mild to moderate asthma, as an add-on to usual care, also looks promising. Everything else in the table is early.
Which breathing exercise works best?
Nobody can say yet. In the one head-to-head trial, breathing exercises and mindfulness lowered anxiety by about the same amount. Slow, gentle breathing without holds has the most support and the fewest known risks, so pick the one you will actually do.
What does "evidence level" mean here?
It is my plain-words summary of how many studies exist, how well they were designed and whether they agree. It is not a formal grade. Nothing in the tracker is rated stronger than early to moderate.
Does a small effect size mean breathing does not work?
No. It means that on average, people in the studies reported somewhat lower scores than comparison groups. Many studies are small and rely on questionnaires, so the true effect may be smaller or larger.
Why do some rows say "abstract only"?
For a few studies I could read only the abstract. I claim nothing beyond what the abstract states, and the sources list below marks them.
Can I download the table?
Yes. The CSV version is at the bottom of this page, and there is a free download link in the builder section above.
Sources and further reading (23)
- Balban MY, Neri E, Kogon MM, et al. Brief structured respiration practices enhance mood and reduce physiological arousal. Cell Reports Medicine, 2023. Full text. PMC9873947
- Riedl EM, Bracklo T, Wimmer K, et al. A pilot, just-in-time ecological momentary intervention study of one-minute cyclic sighing versus box breathing. Anxiety, Stress, & Coping, 2026. doi.org/10.1080/10615806.2026.2659809
- Zaccaro A, et al. How breath-control can change your life: a systematic review on psycho-physiological correlates of slow breathing. Frontiers in Human Neuroscience, 2018. Frontiers
- Fincham GW, Strauss C, Montero-Marin J, Cavanagh K. Effect of breathwork on stress and mental health: a meta-analysis of randomised-controlled trials. Scientific Reports, 2023. PMC9828383
- Steffen PR, Austin T, DeBarros A, Brown T. The impact of resonance frequency breathing on measures of heart rate variability, blood pressure, and mood. Frontiers in Public Health, 2017. PMC5575449
- Goessl VC, Curtiss JE, Hofmann SG. The effect of heart rate variability biofeedback training on stress and anxiety: a meta-analysis. Psychological Medicine, 2017. Abstract only. Cambridge Core
- Pizzoli SFM, et al. A meta-analysis on heart rate variability biofeedback and depressive symptoms. Scientific Reports, 2021. Abstract only. PMC7988005
- Vierra J, Boonla O, Prasertsri P. Effects of sleep deprivation and 4-7-8 breathing control on heart rate variability, blood pressure, blood glucose, and endothelial function in healthy young adults. Physiological Reports, 2022. PMC9277512
- Kirazli G, et al. The effect of 4-7-8 breathing exercise technique on tinnitus handicap, psychological factors, and sleep quality in tinnitus patients. Brain and Behavior, 2026. PMC12895279
- Cheng S, et al. Voluntary slow breathing exercise on cardiovascular parameters in patients with hypertension: a systematic review and meta-analysis. Clinical Cardiology, 2026. PMC13093598
- Garg P, et al. Effect of breathing exercises on blood pressure and heart rate: a systematic review and meta-analysis. International Journal of Cardiology: Cardiovascular Risk and Prevention, 2024. PMC10765252
- Landman GWD, et al. Efficacy of device-guided breathing for hypertension in blinded, randomized, active-controlled trials: a meta-analysis of individual patient data. JAMA Internal Medicine, 2014. Abstract and tables read. JAMA Network
- Santino TA, et al. Breathing exercises for adults with asthma. Cochrane Database of Systematic Reviews, 2020. PMC7096190
- Chelidoni O, et al. Exploring the effects of a brief biofeedback breathing session delivered through the BioBase app in facilitating employee stress recovery. JMIR mHealth and uHealth, 2020. PMC7596654
- You M, et al. Single slow-paced breathing session at six cycles per minute: dose-response relationship on cardiac vagal activity. International Journal of Environmental Research and Public Health, 2021. PMC8656666
- Bentley TGK, et al. Slow-breathing curriculum for stress reduction in high school students: a feasibility pilot. Frontiers in Rehabilitation Sciences, 2022. PMC9397716
- Abera M, et al. Effects of relaxation interventions during pregnancy on maternal mental health, and pregnancy and newborn outcomes. PLOS ONE, 2024. PMC10810490
- Cochrane. Relaxation techniques for pain management in labour (CD009514), plain-language summary. cochrane.org
- Rhee J, et al. Breaking social media fads and uncovering the safety and efficacy of mouth taping: a systematic review. PLOS ONE, 2025. PMC12094774
- Dallam G, Kies B. The effect of nasal breathing versus oral and oronasal breathing during exercise: a review. Journal of Sports Research, 2020. Abstract only. Conscientia Beam
- Bayrak O, Polastri M, Pehlivan E. Effects of nasal and oral breathing on respiratory muscle and brain function: a review. Thoracic Research and Practice, 2025. PMC12047190
- Gerritsen RJS, Band GPH. Breath of Life: The Respiratory Vagal Stimulation Model of Contemplative Activity. Frontiers in Human Neuroscience, 2018. First part read. PMC6189422
- Stanford University. NCT06526585, ClinicalTrials.gov registry record. clinicaltrials.gov
The main table as CSV
Updated 24 October 2026. We re-check posts when new research comes out.
