Does box breathing actually work?

What is box breathing?

Four counts in, four held, four out, four held, repeated. The equal-length sides are where the name comes from, and the technique reached the public through military performance training before it reached clinical research.

That order matters for how you read the evidence. The practice was popular for years before anyone tested it against a comparison, which is a common pattern in this field and a reason to check the numbers rather than the reputation.

What did the trial find?

Real effects, and a disappointing comparison. Yilmaz Balban and colleagues at Stanford randomised 108 people to five minutes daily for 28 days across four arms: mindfulness meditation, cyclic sighing, box breathing, and cyclic hyperventilation.

Box breathing reduced state anxiety by 3.75 points and negative affect by 0.83, both at p below 0.0001. Positive affect rose by 1.84, at p equals 0.026. Those are genuine improvements against baseline.

The comparisons are where it thins out. Box breathing showed no significant difference from mindfulness meditation on anxiety or negative affect. Its respiratory rate change was also not significantly different from meditation. And cyclic sighing, which emphasises prolonged exhalations, was significantly better at increasing positive affect.

How much weight can that carry?

Not much on its own. The box breathing arm had 19 participants, and the authors say directly that the small group sizes reduced their power to compare breathwork techniques against each other.

The study also ran remotely, so adherence was tracked through daily surveys rather than observed, and there was no follow-up beyond the 28 days.

Nineteen people is enough to detect a change from baseline and thin for concluding that one technique matches another. Read the null comparisons as unsettled rather than settled.

Is there other evidence?

A little, and it should be handled carefully. A 2025 randomised trial gave 140 women recovering from mastectomy ten cycles of box breathing alongside standard care, and reported that by the final assessment none of the intervention group remained in the high stress band while 90 per cent of the control group did.

That result is extraordinarily large. Effects of that size in behavioural research usually indicate something about the measurement or the control condition rather than the technique, and this was a single unblinded trial using a self-report questionnaire in one clinical population.

It is worth citing and it should not shift your expectations much. The Stanford trial, with its active comparison and its modest results, is the better guide.

What is worth doing?

Use it if you like the structure. Five minutes daily produced measurable reductions in anxiety, and a technique you will actually repeat beats a better one you abandon.

Consider lengthening the exhale. The arm that performed best in the same trial extended exhalations rather than balancing them, which fits the physiology: the calming shift arrives on the out-breath. The wider breathwork evidence covers that comparison in full.

Drop the retentions if they make you tense. Holding the breath after exhaling is the part people find uncomfortable, and nothing in this evidence shows the holds are doing the work.