Box breathing: how to do it, how many rounds, and what the evidence actually shows

How to do box breathing, step by step

Box breathing is four phases of equal length, repeated: in, hold, out, hold. Cleveland Clinic's guide, from integrative medicine specialist Melissa Young, MD, gives the method in five steps, and it matches the protocols used in the trials below.

1. Breathe out slowly until the lungs feel empty.
2. Breathe in through the nose while counting slowly to four in your head.
3. Hold the breath for a count of four.
4. Breathe out for a count of four.
5. Hold again, with the lungs empty, for a count of four. That is one round; go back to step 2.

Two details carry most of the technique. Every side is the same length: if four seconds is too long for the holds, all four phases shrink together, to three seconds each, rather than the holds alone being cut. And the holds are gentle: Cleveland Clinic's instruction is to stay at your comfort level, breathing gently and not straining, and the 2023 trial told participants to shorten the phases if they had to strain. Breathe through the nose where possible; that trial asked for it too.

How many rounds of box breathing?

The guides say three to four rounds; the trials that measured an effect used five minutes, which at four-second sides is about 19 rounds. Cleveland Clinic's guide says to repeat for three to four rounds and to practise once or twice a day. At 16 seconds a round, four rounds is 64 seconds. That is a fine first try, but it is not the dose anyone tested.

The two trials that put box breathing against a comparison both used five minutes. In Balban and colleagues' 2023 trial in Cell Reports Medicine, participants repeated the pattern "for the entire 5 min", once a day for 28 days. In McAllister and colleagues' 2026 trial in Comprehensive Psychoneuroendocrinology, participants did five minutes of 4-4-4-4 breathing before a stress test and one more minute in the middle of it. Neither paper states a round count; both set a time, so the number of rounds depends on the length of each side.

So "how many rounds" is better answered as "how many minutes". If you want to practise what the trials practised, set five minutes and let the rounds fall where they fall. If five minutes of holds is too much at first, three-to-four rounds is a reasonable place to start and build from, which is how Cleveland Clinic frames it. No study has compared doses, so nothing shows that 19 rounds is better than four.

How long each side should be

Four seconds is a starting point, not a rule; the 2023 trial set each person's side length from a simple breathing test, from 3 to 10 seconds. Before starting, participants took four normal breaths, then one maximal breath in, then breathed out as slowly as they could and timed how long it took to empty the lungs. The study called that the CO2 discard time and used it to pick the length of every phase.

The bands have gaps (20 to 25 seconds and 45 to 50 seconds are not assigned) and the table was the study's own method, not a validated clinical scale. What it shows is useful anyway: the people who designed the largest box breathing trial expected side lengths to differ between people, and told anyone straining at their length to shorten it. A longer box is not a better box; an even, unforced one is the instruction.

What the evidence shows

The evidence that box breathing calms people in the short term is real but limited: a handful of small randomised trials, mostly in healthy adults, with no long-term follow-up. That is the conclusion of the one systematic review of box breathing trials, published in 2026, and it is what the individual trials show when read in full.

The largest trial: box breathing against mindfulness

In Balban and colleagues' month-long trial, box breathing improved mood after each session, but not significantly more than five minutes of mindfulness meditation did. Of 108 participants, 21 were randomised to box breathing, 30 to cyclic sighing, 33 to cyclic hyperventilation with retention and 24 to mindfulness meditation. Each day they rated their mood and anxiety before and after the session. The paper's headline result is about breathwork as a group, "especially the exhale-focused cyclic sighing", which produced a greater improvement in mood than mindfulness (p < 0.05) and was the only practice to cut respiratory rate significantly more than mindfulness did.

For box breathing alone, positive mood rose by 1.84 points a session on average against 1.22 for mindfulness, and the authors describe that difference as not significant. State anxiety fell after sessions in both the box breathing and the mindfulness groups, by similar amounts. The trial was remote, adherence was taken from daily surveys, it ran for four weeks with no follow-up, and the authors say it was "originally intended as an exploratory study" and registered retrospectively. Its control was another practice, mindfulness, not doing nothing.

The stress-test trial: heart rate, not cortisol

McAllister and colleagues' 2026 trial is the clearest test of box breathing before a stressful task, and it found effects on heart rate and one stress marker, not on cortisol or heart rate variability. Sixty-six adults were split between normal breathing, box breathing and a prolonged-exhalation pattern, then gave a speech to a pre-recorded judging panel and did a Stroop task and mental arithmetic, a modified version of the Trier Social Stress Test.

From before the test to after it, heart rate rose 4.4 beats a minute with normal breathing and fell 6.4 with box breathing. Salivary alpha-amylase, a marker of sympathetic nervous system activity, rose 27.7 U/mL with normal breathing and did not change with box breathing. State anxiety rose slightly only in the normal-breathing group, though the groups did not differ significantly right after the test. The authors list the limits: a recorded rather than live panel, a shortened stressor that probably explains why cortisol never rose, no power calculation, and heart rate variability measured only at rest.

The systematic review: limited and fragmented

The 2026 review found eight randomised trials of box breathing, and concluded that more carefully run trials with uniform breathing regimens are needed before wider clinical use can be determined. Dujawara, Ali and Kumari searched PubMed, Scopus, Web of Science and ClinicalTrials.gov to the end of 2025, assessed bias with the Cochrane RoB 2 tool, and summarised the results in words because the trials were too different to pool. They report reduced postpartum pain and some lactation-related improvements in clinical groups, and note that no trial measured lung function or aerobic capacity. One caution on the paper itself: its abstract says eight trials were included and later refers to "the 29 RCTs", which it does not explain.

Breathwork in general

Across all kinds of breathwork, the best pooled estimate is a small-to-medium reduction in self-reported stress. Fincham and colleagues' 2023 meta-analysis in Scientific Reports pooled 12 randomised trials with 785 adults for stress (g = −0.35, 95% CI −0.55 to −0.14), 20 for anxiety (g = −0.32) and 18 for depression (g = −0.40), and judged most trials at moderate risk of bias. Box breathing is one technique inside that family; the pooled figure is not a box breathing result.

What box breathing has not been shown to do

Box breathing has not been shown to lower cortisol, raise heart rate variability, or treat an anxiety disorder, though all three are widely claimed. Cleveland Clinic's 2021 article says regulating your breath can lower the stress hormone cortisol. The one box breathing trial that measured cortisol, McAllister's, found no treatment effect, and the 2026 review summarises the recent evidence as reductions in heart rate, state anxiety and salivary alpha-amylase, "but not HRV or cortisol".

Heart rate variability is the claim most often repeated on wellness pages. Neither trial found it: Balban's showed no differences between any breathwork group and mindfulness for resting heart rate or heart rate variability, and McAllister's found no change over time or between groups. Both were small and short, so this is an absence of evidence rather than proof of no effect, but a page that says box breathing raises heart rate variability is saying more than the trials do.

The technique is often sold as the Navy SEAL method. Whatever its history, both trials described in detail here were run on civilian volunteers, the longer of the two lasted four weeks, and the 2026 review lists short follow-up among the limits of the whole evidence base. Nothing in them tells you what happens after months of practice, or in people with a diagnosed anxiety disorder.

Keeping count of rounds without losing the box

Count the seconds of each side in your head and let something else hold the number of rounds. Counting "one, two, three, four" on every side is already one count; keeping a second running total of rounds on top of it is where people lose their place, usually at the second hold, the one that gets cut short.

One way to split the job: open the breath counter and tap once at the end of each box, as the second hold ends. Leave each count as one whole breath. The Counted figure is then your number of rounds, and Your pace is rounds a minute, taken from the middle value of your recent gaps between taps. At four-second sides it should settle near 3.8 a minute (3.75 rounded); if it creeps towards 5, your box has shrunk to about three seconds a side. The panel is one large key, so you can tap with your eyes closed.

The breath counter does not set the rhythm, deliberately: it counts your taps and does not pace you, which its page explains in a breath counter or a breathing timer. If you would rather follow a moving square, a breathing timer is the right tool; if you would rather keep your own pace and stop at five minutes, a counter is. For a daily practice, a habit counter keeps the streak of days.

When to ask a clinician first

If you have a heart condition, glaucoma, a history of seizures, a lung condition, panic attacks, or are pregnant, ask your doctor before practising breath holds. Those are not guesses: Balban's trial excluded people with self-reported moderate to severe psychiatric or medical conditions "for health and safety reasons", naming heart disease, glaucoma, seizures, pregnancy, psychosis, suicidality, bipolar disorder and substance use disorders. The results above therefore say nothing about those groups.

• Lung conditions such as COPD or asthma. None of the trials in the 2026 review measured lung function, so the evidence is silent here. Use the breathing exercises your respiratory team gives you.
• Panic disorder. The holds are the hard part. In a small 1994 study by Asmundson and Stein in the American Journal of Psychiatry, 23 people with panic disorder held their breath for significantly shorter times than 26 healthy controls. A pattern with no holds, such as the NHS's, may suit better; ask whoever treats you.
• Anxiety under treatment. Box breathing is a short-term tool, not a treatment. Fincham's meta-analysis advises clinicians to consider for each person whether breathwork may worsen the symptoms of certain mental or physical health conditions.

Stop and breathe normally if you feel dizzy, light-headed, short of breath or panicky. Anyone struggling to breathe, or with chest pain, needs medical help, not a breathing exercise.

Box breathing and the other patterns

Box breathing is the only common pattern with two equal holds; the others drop the holds or lengthen the out-breath. In Balban's trial the exhale-focused pattern, cyclic sighing, did best, which is worth knowing if the holds are what you struggle with.

4-7-8 breathing, with one long hold and a longer out-breath, is a separate technique with its own evidence and is not covered here.

What this guide does not cover

This guide explains how box breathing is done, how its rounds and timing work, and what published trials found, as read on 4 November 2026. It is not medical advice, it does not diagnose or treat anything, and it is no substitute for care from a doctor, psychologist or respiratory team. Box breathing is studied as a short-term stress tool in mostly healthy adults; for anxiety, panic, a breathing condition or any symptom that worries you, talk to a clinician.

Simple Online Counter is a free, browser-based set of counters; its breath counter counts breaths or rounds that you tap, times breaths per minute, and keeps every count in your own browser. It does not pace your breathing, assess it, or make any health claim.