Box Breathing: What the Evidence Shows

Four counts in, four hold, four out, four hold

Box breathing is a paced breathing pattern with four equal parts: breathe in for a count of four, hold for four, breathe out for four, and hold for four before starting again. Each cycle takes sixteen seconds, so you end up breathing about four times a minute, considerably slower than the twelve to twenty breaths a minute typical of an adult at rest. The name comes from the four equal sides.

It is one member of a family of slow breathing techniques. Others include resonance or coherence breathing, usually around five to six breaths a minute with inhale-to-exhale ratios that vary, and the physiological sigh, a double inhale followed by a long exhale. These patterns share the basic idea of deliberately changing the breath, but they are not physiologically identical.

Box breathing is a self-regulation skill and may be a useful treatment adjunct. It is not a stand-alone replacement for assessment, psychotherapy or prescribed treatment when distress is persistent. Its practical value is in giving you something concrete and portable to do with your attention and body when arousal is climbing.

Breathing is the one automatic function you can easily take over

Heart rate, digestion and blood pressure run largely without your involvement. Breathing is different. It runs automatically, roughly twenty thousand times a day, yet you can take deliberate control of it at any moment. That makes it an unusually direct handle on the autonomic nervous system, the part of the nervous system that governs the balance between alertness and rest.

When breathing slows below about ten breaths a minute, measurable things can change. A systematic review of fifteen small and varied studies in healthy people reported increases in heart rate variability and respiratory sinus arrhythmia, both related to autonomic regulation. It also reported increased EEG alpha and decreased theta activity. Participants described greater comfort and alertness and less arousal, anxiety and anger. The EEG direction is clear in the review, but translating it into a precise mental state or mechanism would go beyond those data.

Carbon dioxide explains one clinically useful part of the picture. Carbon dioxide is not simply waste. It helps regulate blood vessels in the brain. Hyperventilation means breathing faster or deeper than the body's metabolic needs. When this happens, carbon dioxide can fall, cerebral blood vessels narrow, and blood becomes temporarily more alkaline. This can produce lightheadedness, tingling around the fingers or lips, and a sense of unreality. Those sensations can then frighten someone into breathing harder.

When these symptoms are actually caused by hyperventilation, they usually settle as breathing and carbon dioxide return toward normal. That may be part of why paced breathing helps some people during panic. Not every panic attack involves overbreathing, however, and researchers disagree about how central this mechanism is. A first, severe or unusual episode should not automatically be assumed to be panic, especially with chest pain, fainting, marked breathlessness or new neurological symptoms.

The vagal explanation is a reasonable working model rather than a settled fact. Slow breathing can shift autonomic measures, and vagal pathways are one plausible route. It does not follow that every reported benefit is caused by that mechanism. As the next section shows, trials have had difficulty demonstrating that changing a physiological marker necessarily changes symptoms.

A modest average signal, a strong placebo test, and an acute pilot

A 2023 meta-analysis in Scientific Reports pooled randomised controlled trials of breathwork in adults. Across twelve trials and 785 participants, breathwork was associated with lower self-reported stress than control conditions, with a small to medium standardised effect of g = -0.35. Secondary analyses found g = -0.32 for anxiety across twenty trials and g = -0.40 for depressive symptoms across eighteen trials. The studies covered nonclinical, mental-health and physical-health populations. Most stress studies carried a moderate risk of bias, and the authors urged caution about a miscalibration between hype and evidence.

A 2023 Stanford study in Cell Reports Medicine tested box breathing directly. Participants practised five minutes a day for one month in one of four groups: box breathing, cyclic sighing with prolonged exhales, cyclic hyperventilation with retention, or mindfulness meditation. Combined breathwork improved daily positive affect more than meditation, with cyclic sighing providing the clearest individual signal. Combined breathwork and cyclic sighing also reduced resting respiratory rate relative to meditation.

Two qualifications belong with that result. The trial was not powered to distinguish the three breathing techniques from one another, so it does not establish that cyclic sighing beats box breathing. There were no between-group differences for state anxiety or negative affect, although all groups showed immediate reductions in these outcomes after practice. The study also found no significant change in heart rate variability in any group over the month.

A larger placebo-controlled trial later randomised 400 people to coherent breathing at about five and a half breaths a minute or a convincing sham at twelve breaths a minute. Participants did not know which pattern they had received. Both groups improved on stress, anxiety, depression and wellbeing, with no detectable group-by-time difference for stress, p = 0.765, or the other outcomes. The authors found no measurable benefit of that coherent-breathing protocol over the active breathing placebo.

That result does not identify which non-specific component produced improvement. Quiet time, daily structure, attention, expectation and the act of paced breathing were bundled together in both groups. It therefore supports caution about claims for one exact tempo or neurological mechanism, rather than proving that any one of those common components caused the change.

A 2026 preregistered pilot adds a different piece of evidence. Forty-seven university students used one minute of box breathing, cyclic sighing or a passive control during real-world episodes of anxiety. Across 782 usable anxiety observations, both breathing exercises were associated with greater anxiety reduction five minutes later than the passive control. They did not differ significantly from each other, p = 0.19. Reaction times were slower after both breathing exercises, and the sample was small, young, mostly female and nonclinical. This is evidence for a possible acute effect, not proof of superiority or a durable treatment effect. A free guided cyclic sighing exercise is available on this site with optional spoken prompts.

A 2024 randomised trial points the same way from another angle. Heart rate variability biofeedback, which uses slow paced breathing, successfully raised participants' resting heart rate variability compared with an active control, yet produced no accompanying change on a standard anxiety questionnaire. That trial was small, with forty participants analysed, so it is weak evidence on its own. Alongside the placebo trial it supports a cautious reading: moving a physiological marker does not automatically move symptoms.

A useful adjunct, not a stand-alone replacement for care

Paced breathing may be a useful adjunct, but it should not be presented as a stand-alone replacement for assessment or evidence-based treatment of an anxiety disorder, obsessive compulsive disorder, post-traumatic stress disorder or depression. The pooled trials included a mixture of nonclinical, physical-health and mental-health populations. Their average effects do not by themselves establish treatment efficacy for any one diagnosed disorder.

Some studies examine immediate effects and others examine repeated daily practice. The 2026 pilot found an acute signal after a one-minute exercise, while the longer trials used repeated practice over three or four weeks. Evidence about durability, optimal frequency and long-term clinical benefit remains limited. One coherent-breathing trial included a one-month follow-up, but did not find an advantage over its active placebo.

Nobody should stop or reduce prescribed psychiatric medication because a breathing practice is helping. Stopping antidepressants abruptly can cause a withdrawal reaction that is easily mistaken for relapse, and changes to treatment need a prescribing conversation.

How a technique is used matters. If breathing becomes a rigid ritual that you believe must prevent catastrophe before every meeting, phone call or social event, it may function as a safety behaviour and interfere with exposure learning. Flexible use to settle enough to stay in a situation is different. The same distinction matters in social anxiety and panic disorder.

Slow breathing and forceful breathing are not the same thing

Slow, comfortable paced breathing is low risk for most people. Box breathing adds brief holds, which can provoke air hunger. Forceful techniques such as kapalbhati, breath of fire, and cyclic hyperventilation with retention are a different category because they can lower carbon dioxide sharply and may increase arousal. Advice for one category should not automatically be applied to the others.

Discuss forceful hyperventilation practices with a clinician first if you have uncontrolled hypertension, heart or lung disease, epilepsy, a history of fainting, glaucoma, a hernia, an active peptic or duodenal ulcer, recent eye or abdominal surgery, or if you are pregnant. Hyperventilation can activate epileptiform activity, especially in absence epilepsy. Reflux alone is not listed here as a blanket contraindication because I could not verify adequate support for that claim.

Psychosis, bipolar disorder and current suicidal thoughts are not established direct contraindications to a brief breathing exercise. The prudent clinical distinction is intensity and current stability. Avoid unsupervised, prolonged or highly activating breathwork during acute mania, psychosis, severe agitation or suicidal crisis, and do not use it instead of urgent clinical care. This precaution is partly extrapolated from adverse-event reports involving intensive meditation and yoga, rather than direct trials of brief forceful breathing.

One caution is non-negotiable. Never practise breath holds or forceful breathing in or near water, and never before or during swimming or diving. Hyperventilating before a breath hold can delay the carbon-dioxide-driven urge to breathe while oxygen continues to fall. Loss of consciousness can then occur with little warning. Fatal and nonfatal cases have been documented in strong, experienced swimmers.

The breath holds in box breathing deserve a smaller caution of their own. For some people with panic, holding the breath itself provokes air hunger and increases fear. If that happens, the sensible adjustment is not to push through it. Drop the holds and simply breathe in for four and out for six, which many people find easier to tolerate.

Stop and return to normal breathing if you feel faint, unusually dizzy, or notice tingling spreading through your hands or face. Practise seated rather than standing, and never while driving. Seek urgent medical assessment for a first, severe or unusual episode, especially with chest pain, fainting, marked breathlessness, one-sided weakness, difficulty speaking or persistent symptoms.

Short and regular beats long and occasional

One to five comfortable minutes is a reasonable starting point. Studies have tested sessions ranging from about one minute during acute anxiety to five or ten minutes of daily practice. Sit with your shoulders loose, breathe through your nose if comfortable, and avoid taking unusually large breaths. Counting silently gives your attention somewhere to rest.

If four counts feels like a strain, shorten everything to three. Straining defeats the purpose. No single rhythm has established overall superiority, so tolerability and consistency matter more than forcing an exact count. If the holds bother you, drop them and use a comfortable slow inhale with a slightly longer exhale.

Practising while calm can make the sequence more familiar, but it can also be tried during rising anxiety if it feels tolerable. It is not a test of willpower. Stop if the exercise increases fear, breathlessness or dizziness.

If you would like a paced timer to follow rather than counting alone, this site has a free guided box breathing exercise, along with other relaxation exercises including progressive muscle relaxation. Nothing is recorded and no sign-up is required.