Breathing exercises for sleep: what's tested and what's not

Slow breathing is the most recommended sleep technique in the world and one of the few with solid physiology behind it. It’s also the one where the best-designed trial found nothing. Both things are true, and the gap between them is the interesting part.

The physiology is real

This isn’t in dispute. A systematic review of the psychophysiology of slow breathing (defined as fewer than about 10 breaths per minute) found consistent effects: increased heart-rate variability and respiratory sinus arrhythmia, indicating a shift toward parasympathetic dominance; increased alpha and decreased theta power on EEG; and subjective reports of greater comfort, relaxation, and pleasantness alongside reduced anxiety, anger, and confusion (Zaccaro 2018).

Breathing is the only autonomic function you can drive voluntarily, and driving it slowly reliably moves the rest of the system. That much is settled.

The sleep evidence is thinner than you’d expect

Settled physiology is not the same as a settled sleep benefit, and here the studies get small.

Fourteen self-reported insomniacs and fourteen good sleepers practiced 20 minutes of slow paced breathing before bed, with polysomnography measuring the night. The insomnia group showed shorter sleep-onset latency, fewer awakenings, less time awake during the night, and better sleep efficiency (Tsai 2015). Encouraging — and fourteen people.

A pilot trial gave 20 participants two consecutive nights of polysomnography, one intervention night and one control night. The intervention was either 30 minutes of pre-sleep slow breathing or music listening, so each condition rested on roughly half the sample. Slow breathing increased central delta power during deep sleep. But the authors’ own summary is the honest one: while some indices pointed toward improvement, neither condition had robust effects on sleep quality (Kuula 2020).

The result the breathwork industry skips

Now the big one, and it’s the reason this post exists.

Researchers randomized 400 people to four weeks of daily coherent breathing (the classic slow, even, roughly six-breaths-a-minute pattern) or to a carefully built attention-matched placebo: the same daily commitment and the same equal inhale-to-exhale ratio, but at 12 breaths per minute, comfortably above the threshold where the slow-breathing physiology kicks in.

Everyone improved. Nobody improved more. There was no significant group-by-time interaction and no main effect of group on stress, anxiety, depression, or wellbeing. The authors found “no measurable effect of coherent breathing over and above a well-designed breathwork placebo” (Fincham 2023).

This is the largest and most tightly controlled breathwork trial we found. It measured mental health rather than sleep, but it suggests much of what people get from a breathing practice comes from doing a calm, structured thing every day, not from the specific breath count.

That’s not nothing. A ritual that reliably makes you feel calmer is worth having, and it’s free. But it does mean the confident claims about particular ratios are running well ahead of the evidence.

Which brings us to 4-7-8

The most famous sleep-breathing pattern in the world (inhale for four, hold for seven, exhale for eight) has almost no direct research behind it. Searching the medical literature for 4-7-8 breathing and sleep returns a small handful of papers: one in tinnitus patients, one in post-surgical symptom management, and a physiological study measuring heart rate and blood pressure in sleep-deprived people. We could not find a single trial testing whether it helps people with sleep problems sleep.

That doesn’t mean it doesn’t work. Long exhales and a slow overall rate are exactly the ingredients the physiology review supports, so 4-7-8 is a reasonable thing to be doing. It just hasn’t earned the certainty it’s sold with.

What did beat a comparison group

One well-run study found a specific pattern outperforming something real. 108 people were randomized to five minutes a day for a month of one of three breathwork exercises or of mindfulness meditation. Cyclic sighing — a double inhale through the nose followed by a long, slow exhale through the mouth — produced greater improvement in mood and a greater reduction in respiratory rate than meditation did (Balban 2023).

Read it carefully, though. The arms were uneven (24 in the meditation group against 84 across the three breathwork groups), the authors flag limited statistical power and no follow-up beyond four weeks, and the comparison was mindfulness rather than a breathwork placebo. As Fincham showed, that is the comparison that matters.

And this study did track sleep, by wearable and by a validated sleep-impairment scale. Nothing moved: no significant change in sleep hours, sleep efficiency, or sleep score, within either group or between them. The mood result is real. It did not reach the night.

What to actually do

The through-line across everything above is slow, and specifically long exhales. Almost every technique that works is a different packaging of the same two ideas.

• Aim for roughly six breaths a minute. In and out over about ten seconds total. That’s the rate the physiology consistently likes.
• Make the exhale longer than the inhale. The parasympathetic shift lives on the out-breath.
• Five to twenty minutes is the tested range. Five moved mood in the daily-practice trial; twenty was what the pre-sleep polysomnography study used.
• Don’t agonize over the pattern. 4-7-8, box breathing, coherent breathing, cyclic sighing — the trial with a real placebo suggests the specific numbers matter less than the doing.
• Practice it in the day, not just at 2am. Every trial here involved daily practice. A technique you’ve rehearsed is one you can deploy when you’re wired.
• If it makes you dizzy or anxious, stop. Breath holds and forceful patterns don’t suit everyone, and there’s no prize for pushing through.

If you’d rather be led through it

Doing this from memory in the dark is harder than it sounds: the counting itself becomes something to be alert about. The breathing pacer on this site handles the count for 4-7-8, box, and coherent breathing, free, in the browser. Slumber has guided breathing and muscle-relaxation exercises alongside its sleep meditations and stories, so the pacing is someone else’s job.

---

Sources: Zaccaro et al. 2018 · Tsai et al. 2015 · Kuula et al. 2020 · Fincham et al. 2023 · Balban et al. 2023. More on how the techniques compare. This is general information, not medical advice — persistent insomnia is worth taking to a doctor.