Yoga is safe for most people, and serious shoulder injury is uncommonStrong evidence
In trials, yoga was about as safe as usual care and other exercise, and across 9 observational studies most problems were mild and short-lived. A German survey of 1,702 practitioners found 0.60 acute injuries per 1,000 hours of practice, which its authors judged "as safe or safer when compared to other exercise types", with hand-, shoulder- and headstands the practices most often linked with acute problems.
No study gives a rate of shoulder injury in yoga. In a year-long study of 354 practitioners, 10.7% had new pain caused by yoga, and the university's summary of the study says most of it was in the shoulder, elbow, wrist and hand, possibly from downward dog and similar poses (the paper's abstract, which we read, doesn't give the split). In US emergency department data from 2004 to 2023, the shoulder was 9.0% of yoga injuries (from the study's abstract); an earlier study counted the shoulder inside the trunk, so it gives no shoulder figure.
Sources 1, 2, 3, 4, 5, 6, 7
What yoga asks of the shoulderEmerging evidence
Yoga takes the arms overhead and down again in every sun salutation (in a motion study of 10 practitioners, the elbow bent to about 116 degrees in the eight-limb pose, close to the bottom of chaturanga), puts weight on the hands in plank, chaturanga, the dogs, side plank and arm balances, takes the arms behind the back in binds and across the body in eagle and cow face, and loads the shoulders at full overhead reach in wheel, forearm stand and headstand.
In a force-plate study of 20 adults, plank, chaturanga, downward and upward dog and dolphin put a peak of 0.71 to 1.08 times body weight through the two hands, crow 1.05, and side plank 0.67 through one arm. The authors note that "the upper extremities are not designed to absorb weight as are the lower extremities". Some poses work the shoulder-blade muscles well: in 15 static poses, locust with the arms forward gave the highest trapezius activity.
Sources 8, 9, 10, 11
Who gets it, and what it usually isEmerging evidence
People who step up the number of vinyasas, move to a faster or stronger class, or start arm balances and inversions (our reading of the load studies). People from mid-life, when rotator cuff changes become common anyway: from 9.7% at 20 and under to 62% at 80 and over, many without pain. Very flexible people, whose shoulders can slip: the NHS lists "Shoulder instability, sometimes because of hypermobility". And people whose shoulder was already sore, who notice aches after a class more often.
In a specialist clinic's 89 yoga injuries, 6 (6.7%) were rotator cuff injuries, and imaged yoga injuries at a large centre included tears of a rotator cuff tendon and of the shoulder's labrum. Teachers and clinicians link shoulder injuries with downward dog and plank variations, including chaturanga and side plank, which is belief rather than a measured rate. Most shoulder pain in class is rotator cuff related shoulder pain, an umbrella for impingement, tendon pain and cuff tears.
Sources 12, 13, 14, 15, 16, 17, 18
Rare but serious: the heart, a fall and a squeezed nerveEmerging evidence
Shoulder or arm pain with chest pain, breathlessness, sweating or feeling sick can be the heart, not the shoulder: the NHS heart attack page says to call 999. A fall from an arm balance or a headstand can dislocate or break the shoulder. And one case describes a weak wrist after a woman with an inherited nerve condition tried crow pose again and again for a month, from a nerve squeezed in the upper arm; she recovered almost completely over three months. These are rare: they show what the warning signs are for, not a risk.
Sources 19, 20
When a scan mattersStrong evidence
Most shoulder pain is diagnosed from the story and an examination: a review of 26 guidelines found that magnetic resonance imaging (MRI) is usually not recommended for early shoulder pain, and rotator cuff changes on scans become common with age, many without pain.
A scan or X-ray matters after a fall with a shoulder that looks out of place or an arm you can't lift, for sudden weakness after an injury, when an acute tear can need early surgical review, and when exercise hasn't helped after about 3 months.
Sources 21, 15
How it usually goesStrong evidence
The NHS says "You usually need to do these things for 2 weeks before shoulder pain starts to ease" and "It can take 6 months or longer to recover from shoulder pain." NHS inform says new pain or a flare-up "should begin to settle within 6 weeks".
In the GRASP trial of 708 people with rotator cuff disorders, both progressive exercise and one session of best practice advice with a home programme improved a lot over a year, with no difference between them. No one can promise a date.
Sources 16, 22, 23