Breathwork boosts your immune system

Most studies of breathing practices and the immune system measured inflammation, and what they found was a dampening rather than a boost: breathing exercises lowered signs of inflammation. One trial has counted infections. Among 250 hospital workers in India exposed to COVID-19 in 2020, one of 123 who practised yogic breathing twice a day for four weeks tested positive, compared with nine of 127 who kept to their usual routine. People were placed in the two groups in turn, not at random, the comparison group did no other practice, and people who missed sessions were left out of the count, so with ten infections in all this is an early result that has not been repeated. The trial was funded by India's national yoga institute, whose director was one of its authors.

That trial compared breathing only with doing nothing extra, and no trial has tested breathing practices as a replacement for vaccination or for any treatment. The fast, breath-holding style used in the best-known immune trials carries risks slow breathing does not: fainting, which can be fatal in or near water, and seizures, since brief deliberate hyperventilation is used in clinics to bring them on. Reviews of these styles describe brief rises in blood pressure and advise specific caution for people who are pregnant or who have epilepsy, panic disorder or psychosis.

In two Dutch trials, healthy young men trained in a fast breathing method developed by Wim Hof, built on repeated deep, rapid breaths followed by breath holds, had inflammatory signals roughly half as high as untrained men after a deliberate injection of a bacterial toxin, and reported fewer flu-like symptoms from the toxin. In the first, with 24 men, the training was led by Wim Hof over ten days and included ice-water immersion, time in the snow and meditation alongside the breathing, so the breathing could not be separated from the rest. The second, in 2022, separated the parts among 48 men: cold exposure training alone did not change the response, the breathing exercise alone raised adrenaline and shifted the response toward anti-inflammatory and away from pro-inflammatory signals, and the two together shifted it most. A 2024 review of the method's trials described the same pattern and rated every trial at very high risk of bias.

The method has since been tried in people with an immune-related disease. In a 2026 pilot trial in Slovakia, 60 adults with multiple sclerosis, most of them taking their usual medicines, were randomly assigned to 12 weeks of the Wim Hof method (breathing sessions, dips in a cold river and visualisation), a dance and diet programme, or no change, and 43 finished. Two inflammatory signals fell in both active groups compared with the group that changed nothing, a third fell only with the Wim Hof method and a fourth only with dance and diet; others, including interleukin-6, did not change. Blood markers of nerve damage did not change, 8 of the 20 people given the Wim Hof method dropped out, and the trial did not measure relapses, disability or infections. The researchers behind the toxin trials suggested testing the approach in autoimmune disease; no trial has tested it in place of treatment.

Slow breathing has been tested less. In 46 adults in hospital in Germany with moderate COVID pneumonia, 20 minutes of slow breathing three times a day alongside standard care was followed by lower levels of interleukin-6, an inflammatory signal, than standard care alone; three other inflammation measures did not differ, and the trial did not report whether patients recovered faster. In 55 adults with panic disorder, four weeks of slow breathing with heart rate biofeedback lowered one of the two inflammatory signals measured, tumour necrosis factor, compared with a sham version of the training; interleukin-6 did not change.

Two other sources bear on infections. In Iran, 110 women with COVID-19 who were not in hospital were randomly assigned to 20 minutes a day of yogic breathing plus 20 minutes of meditation for six days, or to no contact; by day six the practice group reported milder headache, muscle pain, cough, breathlessness and weakness, but nobody was blinded, the researcher who taught the practice also collected the symptom reports by phone, the breathing could not be separated from the meditation, and the trial measured symptoms, not infections caught, immune markers or time to recovery. The other source is an online survey in Poland of people who practised a programme combining cold-water swimming and breathwork. They reported shorter colds and other upper respiratory infections over the previous autumn and winter than people who only swam in cold water or did neither; the groups chose their own habits, answered from memory and were not asked about exercise or diet, so the survey shows an association, not an effect. US federal information on relaxation techniques, which include slow, deep breathing, describes research on blood pressure, anxiety, pain, sleep and hot flashes and does not mention immune function or infections; it describes the techniques as generally safe for healthy people, with rare reports of worsened symptoms in people with epilepsy or certain psychiatric conditions. All the trials described here were in adults, and the toxin trials included only healthy young men.