Prāṇāyāma — breath practice, described accurately

What it is for

In Patañjali's scheme, prāṇāyāma sits fourth, after posture and before the withdrawal of the senses. Its stated function is to remove what obscures clarity and make the mind fit for concentration.

That is worth holding onto, because most contemporary teaching presents it as a relaxation technique. Calm is a common effect. It is not what the text says the practice is for, and several of the classical techniques are stimulating rather than calming.

The main techniques

Nāḍī śodhana — alternate nostril. Inhale through one nostril, exhale through the other, alternating. The standard beginning practice, low-risk without retention, and the one most widely taught.

Ujjāyī — a slight constriction at the throat producing an audible breath, used to lengthen and smooth the cycle. Familiar to anyone who has done a vinyasa class.

Bhrāmarī — the humming breath, exhaling with a bee-like sound and the ears lightly closed. Reliably settling. It is often taught with the fingers pressing the eyeballs, which is the one thing anyone with glaucoma or retinal disease should not do; the humming does not require it.

Kapālabhāti — sharp forceful exhalations with passive inhalation. Vigorous; classified in the older texts as a cleansing practice rather than as prāṇāyāma proper.

Bhastrikā — bellows breath, forceful in both directions. The most demanding of the common practices and the one most often taught carelessly.

Śītalī and śītkārī — cooling breaths drawn over the tongue.

Kumbhaka — retention, after inhalation or exhalation. This is where the classical literature places the real work of prāṇāyāma, and also where it places its warnings.

What the research supports

Slow-breathing practices — roughly six breaths a minute — have been studied reasonably well, and the three main findings do not stand at the same strength.

Heart-rate variability is the clearest: a 2022 review pooling 223 studies found vagally-mediated HRV rose during practice and afterwards. Stress and anxiety improve by a small-to-medium margin, though most of those trials carry a moderate risk of bias.

Blood pressure is contested, and this page does not claim it. Pooled trials show falls of roughly five to seven mmHg systolic. But when the analysis is restricted to trials that were blinded and compared against an active control rather than against nothing, the effect disappears — a 2014 individual-patient meta-analysis in JAMA Internal Medicine found 2.2 mmHg in favour of the control group. Both results are in the sources below. Reporting only the first would not be summarising, it would be choosing.

Beyond that, the picture thins quickly. Trials of specific techniques for specific conditions are mostly small, short and of variable quality, and the literature has the same positive-result bias found across yoga research.

What is not supported by that literature — and is often asserted — is that particular techniques detoxify, oxygenate the blood beyond normal levels, or alter the course of serious disease.

The safety section that classes skip

The haṭha texts are blunter about this than most modern classes are. The Haṭha Yoga Pradīpikā requires a teacher (2.1), states that badly practised breath kills the practitioner (2.15), and names what goes wrong — hiccup, breathlessness, cough, and pain in the head, ears and eyes (2.17). That warning is part of the tradition, not a modern caveat bolted onto it.

Never practise any of this in water. Hyperventilating before going under removes the urge to breathe without adding useful oxygen, so a practitioner loses consciousness with no warning. This is the one item on this page with deaths behind it.

In modern terms, the specific cautions are:

• Forceful practices and retention in uncontrolled high blood pressure, glaucoma or retinal disease, epilepsy, and pregnancy.
• Forceful practices if you have had a collapsed lung, or have bullous or emphysematous lung disease. Kapālabhāti has a published case report of causing a pneumothorax in an otherwise healthy 29-year-old.
• Hyperventilation techniques can cause dizziness, tetany and fainting — practise seated, never standing.
• Recent surgery to the abdomen or chest, and active respiratory infection.
• Intensive practice can surface anxiety and dissociation in susceptible people. A 2020 review of 83 studies put the overall adverse-event rate at about 8%, in people with and without any prior mental health history.

Sudden chest pain or breathlessness during or after forceful breathing is a reason to go to an emergency department, not a reason to persevere.

None of that is a reason not to practise. It is a reason to start with the mild techniques, to build slowly, and to treat "more is better" as the error it is.

This page describes a tradition and what research has found. It is not medical advice, and nothing here is a reason to stop a treatment a doctor has prescribed.

How this changes across traditions