The science of yoga is the scientific basis of modern yoga as physical exercise in human sciences such as anatomy, physiology, and psychology. Yoga's effects are to some extent shared with other forms of exercise,^{[O 1]} though it differs in the amount of stretching involved, and because of its frequent use of long holds and relaxation, in its ability to reduce stress. Yoga is here treated separately from meditation, which has effects of its own, though yoga and meditation are combined in some schools of yoga.

Yoga has been studied scientifically since the 19th-century physiology experiments of N. C. Paul. The early 20th-century pioneers Yogendra and Kuvalayananda both set up institutes to study yoga systematically.

Yoga helps to maintain bone strength, joint mobility, and joint stability. It improves posture, muscle strength, coordination, and confidence, in turn reducing the risk of injury and bone fracture. As it is generally slow and conducted with awareness, it may be safer than many other sports; but some postures such as headstand, shoulderstand, and lotus position have been reported as causes of injury.

Yoga is also used directly as therapy, especially for psychological conditions such as post-traumatic stress disorder, but the evidence for this remains weak. Yoga has sometimes been marketed with pseudoscientific claims for specific benefits, when it may be no better than other forms of exercise in those cases;^{[O 1]} and some claims for its effects on particular organs, such as that forward bends eject toxins from the liver, are entirely unfounded. Reviewers have noted the need for more high-quality studies of yoga's effects.

In the 19th century, the Bengali physician N. C. Paul began the study of the physiology of yoga with his 1851 book Treatise on Yoga Philosophy, noting that yoga can raise carbon dioxide levels in the blood (hypercapnia).^[2][3][4]

Early in the 20th century, two pioneers of yoga as exercise in India, Yogendra and Kuvalayananda, worked to make Haṭha yoga acceptable, seeking scientific evidence for the health benefits of yoga postures (asanas) and yoga breathing (pranayama). In 1918, Yogendra founded The Yoga Institute to carry out research on yoga, hoping that a gloss of science would make yoga more acceptable in the West.^{[P 1][O 2][5]} Yogendra expressed his intentions in books such as his 1928 Yoga Asanas Simplified^[6] and his 1931 Yoga Personal Hygiene.^[7] In 1924, Kuvalayananda founded the Kaivalyadhama Health and Yoga Research Center, combining asanas with gymnastics, and like Yogendra seeking a scientific and medical basis for yogic practices.^[8][9]

In 1937, the Yale physiologist K. T. Behanan published his book Yoga: A Scientific Evaluation, reporting that a form of pranayama, Ujjayi ("Victorious breath"), performed at the slow rate of 28 breaths in 22 minutes, could create a deeply relaxed state that he called "an extremely pleasant feeling of quietude",^[10] accompanied by a marked slowing of mental performance on tests such as mental sums, recognising colours and solving simple puzzles. The science journalist William Broad notes that this finding contradicted the image of yoga as conferring special powers.^[11][10]

In 1970 Christopher Hills, his son John, and Kevin Kingsland organized the 'World Conference on Scientific Yoga' in New Delhi, bringing 50 Western scientists together with 800 of India's leading swamis, yogis and lamas to discuss their research and establish a network for the creation of a World Yoga University.^[12]

Yoga as exercise is defined by Merriam-Webster as "a system of physical postures, breathing techniques, and sometimes meditation derived from [traditional] Yoga but often practiced independently especially in Western cultures to promote physical and emotional well-being".^{[O 3]}

The science journalist William Broad notes that yoga has "wide health benefits",^[13] and defines the scope of the science of yoga as to "better understand what yoga can do and better understand what yoga can be".^[14] He distinguishes "the modern variety" which is his subject from the Haṭha yoga that formed "in medieval times".^{[14][lower-alpha 1]} Denise Rankin-Box, editor of Complementary Therapies in Clinical Practice, one of several Elsevier journals that publish papers on the effects of yoga (among other matters), offers the definition "research addressing the impact of yoga on health and wellbeing."^{[O 5]} Ann Swanson, an educator and yoga therapist, writes that "scientific principles and evidence have demystified so much of the practice" of yoga;^[15] her book on the Science of Yoga is principally about the anatomy of yoga asanas,^[16] with a chapter on the relationships of the body's systems (anatomy and physiology) to yoga.^[17] Psychiatric researchers such as Michaela Pascoe have addressed the effect of yoga on measures of psychological stress and depression.^{[P 2][P 3]}

Broad notes the "diffuse nature of the existing science"^[14] with pieces of the metaphorical jigsaw puzzle of scientific knowledge of what yoga actually achieves held in many laboratories around the world. The picture is, Broad writes, confused by the "predatory behaviour"^[18] of commercial ventures intent upon promoting themselves;^[18] but is being clarified by the American National Institutes of Health, which began funding scientific research into yoga in 1998, leading to reliable reports of studies of yoga's effects on different conditions.^[19]

There is strong scientific evidence that yoga is a safe and effective additional therapy for people with low back pain.^{[P 4]} On its own, yoga is about as effective as other active treatments for back pain, and more effective at reducing pain and improving function than non-exercise treatments.^[37]

Yoga is useful to some extent for psychological conditions such as stress and depression,^{[P 2][P 3]} but despite repeated attempts, little or no evidence for benefit for other specific medical conditions. Much of the research on the therapeutic use of yoga, including for depression, has been in the form of preliminary studies or clinical trials of low methodological quality, suffering from small sample sizes, inadequate control and blinding, lack of randomization, and high risk of bias.^{[P 11][P 12]} For example, study of trauma-sensitive yoga has been hampered by weak methodology.^{[P 13]}

The neurologist and sceptic Steven Novella wrote in 2013 that "Yoga .. fits into a more general phenomenon of marketing a specific intervention as if it has specific benefits, when in fact it only has generic benefits" associated with physical exercise.^{[O 1]} Novella points out that yoga also has a spiritual side, so claims made for it can mix science with "a liberal dose of pure pseudoscience and mysticism."^{[O 1]} He illustrates this by quoting unfounded claims such as that a forward bend squeezes the pancreas and liver, ejecting toxins, and that stretching the lower back is calming because emotional stress accumulates in the lower back muscles. Novella states that "None of those specific claims is based in reality."^{[O 1]}