Every major guideline puts exercise firstStrong evidence
NICE's core treatments are therapeutic exercise and weight management, with information and support. OARSI (the Osteoarthritis Research Society International) makes structured land-based exercise a core treatment for knee, hip and widespread osteoarthritis. The American College of Rheumatology strongly recommends exercise, weight loss if needed, tai chi and self-management programmes, and EULAR (the European Alliance of Associations for Rheumatology) includes exercise in its knee, hip and hand recommendations.
In 2018, EULAR concluded that the activity guidelines for everyone, covering aerobic, strength, flexibility and balance work, are feasible and safe for people with hip or knee osteoarthritis and inflammatory arthritis, and that promoting them should be part of standard care. Its 2025 update, published in 2026, keeps activity promotion as standard care and puts more weight on sitting less and on support to change habits, including through technology.
Sources 1, 2, 3, 27, 28, 4, 29
Knee: a modest average benefitMixed evidence
The 2024 Cochrane review (139 trials, 12,468 people) found pain about 9 points better on a 0 to 100 scale than placebo or attention control (low certainty), and about 13 points better than usual care; function was about 11 to 13 points better (moderate certainty). The authors judged these gains of uncertain clinical importance, because they didn't clearly reach the change most people notice (12 points for pain, 13 for function). Types of exercise didn't differ, and exercise may increase side effects (low certainty).
Source 5
Hip: a small but reliable benefitStrong evidence
In 9 trials (549 people), exercise reduced hip pain by about 8 points and improved function by about 7 points on 0 to 100 scales, with high-quality evidence: about 1 in 6 people improved who otherwise wouldn't have. The benefit lasted 3 to 6 months after supervised sessions ended.
Source 6
Hands: small and short-livedEmerging evidence
In 5 trials (381 people), hand exercise reduced pain by about half a point on a 0 to 10 scale, with small gains in function and stiffness. The evidence was low quality, and the benefit had faded at 6 and 12 months.
Source 7
Shoulder, elbow, ankle and neck: hardly any trialsMixed evidence
There are almost no exercise trials for osteoarthritis in these joints. The advice is borrowed from NICE's guidance, which covers osteoarthritis in any joint, and our joint guides say so.
Source 1
Water exerciseEmerging evidence
In 13 trials (1,190 people, average age 68), exercise in water gave a small short-term improvement in pain and disability, about 5 points on a 0 to 100 scale, with no serious side effects. OARSI recommends it for knee osteoarthritis but not for the hip or for many joints at once.
Sources 16, 2
Group programmes: GLA:D and ESCAPE-painEmerging evidence
In Denmark's GLA:D programme (Good Life with osteoArthritis in Denmark), 9,825 people did two education sessions and 12 supervised exercise sessions over 6 weeks. Knee pain fell from 48 to 34 out of 100 and hip pain from 47 to 36 at 3 months, it held at a year, and fewer people took painkillers; there was no comparison group, so the true effect is smaller.
In the UK, a trial of 418 people with long-term knee pain found that ESCAPE-pain, a group programme of exercise and self-management, improved function a little more than usual GP care. More than 20,000 people have now done the programme.
Sources 30, 31, 32
Which type works best?Mixed evidence
The reviews disagree. The Cochrane review found no difference between types for the knee. A meta-regression of 48 knee trials found aerobic, strengthening and functional exercise similar, with thigh-focused programmes and supervised sessions at least 3 times a week doing better. A network analysis of 103 trials ranked aerobic and mind-body exercise (tai chi, yoga) highest for pain, and an earlier one of 60 trials favoured combining strength, flexibility and aerobic work.
Tai chi twice a week for 12 weeks matched physiotherapy in a trial of 204 people with knee osteoarthritis, and the American College of Rheumatology strongly recommends it.
Sources 5, 33, 14, 34, 15, 3
Safe over the long termStrong evidence
Across 49 studies of exercise lasting 3 to 30 months in older adults with knee pain, there were no serious side effects, no increase in pain, no worsening on scans and no increase in knee replacements. In the 66 trials behind EULAR's recommendations, most trials that reported side effects found none, and the rest found only minor ones such as temporary joint or muscle pain. Loading exercise didn't seem to harm knee cartilage, though that evidence is low quality, and NICE's experts wanted to reassure people that exercise is not harmful to osteoarthritic joints.
Sources 9, 4, 10, 8