Exercise for knee arthritis: a modest benefitStrong evidence
In the 2024 Cochrane review (139 trials, 12,468 people), exercise improved pain by 8.7 points out of 100 against placebo or attention control and by 13.1 against usual care or no treatment (both low certainty), and function by about 11 to 13 points (moderate certainty). The authors took 12 points for pain and 13 for function as a noticeable change, so they called the benefits of uncertain clinical importance. They found no differences between types of exercise, and side effects were more common than with usual care (low certainty). The 2015 version, with 44 trials, had rated its evidence higher.
Sources 6, 22
Pilates against no exerciseEmerging evidence
In the 2025 review of 11 trials (476 people aged 43 to 68, from Iran, India, Brazil, Pakistan, Nigeria, Egypt and Turkey), Pilates eased pain more than no exercise by a large but uncertain amount (3 trials, 66 people, low certainty). A second 2025 review of 8 trials (322 people) agreed, with signs that small trials with negative results may be missing.
Sources 4, 5
Pilates against other exerciseEmerging evidence
Against conventional exercise, there was no clear difference in pain or function (5 trials, 210 people, very low certainty). A 2022 review of 7 studies in older adults with joint and muscle conditions found Pilates as effective as other exercise, with good attendance at group classes and poorer practice at home.
Sources 4, 5, 23
Rankings that put Pilates firstMixed evidence
Two network meta-analyses ranked Pilates top for pain or for function, but each rested on a couple of small Pilates trials and indirect comparisons, and one of them warned that its rankings should be read cautiously. The largest, in the BMJ (217 trials), found aerobic exercise likely the most helpful overall, with mind-body exercise such as tai chi, yoga and Pilates giving a large short-term gain in function. Rankings like these don't show that Pilates is the best exercise.
Sources 24, 25, 26
The trials one by oneMixed evidence
All are small, short and from single centres. In older women in Brazil, knee scores improved with mat Pilates twice a week but not without exercise (the abstract gives no test between the groups). In 40 women in Pakistan, Pilates beat isometric exercise with heat and TENS (a kind of electrical pain relief); in 41 people in Iran, it beat conventional exercise; in 30 Iranian women, suspension training beat Pilates. Adding pain education before Pilates eased fear of movement but not pain (54 adults). In 20 Italian women, in a study that wasn't randomised, pain fell 33% with Pilates and 83% with Gyrokinesis. Only 2 of the 11 trials in the 2025 review reported side effects.
Sources 27, 28, 29, 30, 31, 32, 4
Strength, structure and supervisionEmerging evidence
In one analysis of 48 trials, programmes aimed at the thigh muscles (quadriceps), and supervised exercise 3 times a week, reduced pain more, though the 2024 Cochrane review found no link with the number of sessions. In Denmark, the GLA:D programme (education plus 12 supervised exercise sessions over 6 weeks) went with pain about 12 to 14 points lower at 3 and 12 months in 9,825 people, without a comparison group, and the UK's ESCAPE-knee pain programme improved function more than usual care in 418 people.
Sources 7, 6, 20, 33
Exercise first, then surgery if neededStrong evidence
In 100 people eligible for a knee replacement, about three-quarters of those given 12 weeks of exercise and other non-surgical care hadn't had a replacement a year later. Surgery improved symptoms more but brought more serious complications. NICE puts exercise first and suggests referral for a replacement when symptoms substantially affect quality of life and non-surgical care hasn't worked.
Sources 34, 1
Guidelines, tai chi and yogaStrong evidence
NICE, OARSI, the American College of Rheumatology and EULAR all make exercise a core treatment for knee osteoarthritis, and none names Pilates. The American College of Rheumatology found too little evidence to recommend one form of exercise over another; it strongly recommends tai chi and conditionally recommends yoga for the knee. In 204 people, tai chi twice a week did as well as physical therapy. No trial has compared Pilates with yoga or tai chi.
Sources 1, 35, 36, 37, 38