Design: Systematic review with Bayesian network meta-analysis of randomised trials
Population: 35 trials, 3,808 people with knee osteoarthritis
Primary outcome: Pain, function and quality of life
Effect: Short-term combined exercise: pain SMD -0.94, function -0.59 (low certainty)

A Bayesian network meta-analysis compared aerobic, resistance and combined exercise across 35 randomised trials in 3,808 people with knee osteoarthritis, separating short-term (up to 12 weeks) and longer programmes. It was published online in July 2026.

Short-term combined aerobic and resistance exercise ranked highest for pain (standardised mean difference -0.94), function (-0.59) and quality of life (0.66). Longer-term aerobic exercise also improved function. All exercise types helped. Certainty was low to very low.

The finding supports what most guidelines already say: exercise is core treatment for knee osteoarthritis. What it adds is a nudge to include both strengthening and aerobic components rather than one alone.

The low certainty reflects small, varied trials; the ranking between exercise types is less reliable than the overall message that exercise works.

Prescribe exercise for every patient with knee osteoarthritis, before and alongside other treatment.
Include both strengthening and aerobic work, such as walking or cycling.
A 12-week supervised block showed the clearest pain benefit.
Plan how the patient will keep exercising after the programme ends.

Why it matters

The most effective knee osteoarthritis treatment is still one you prescribe, not one you inject.

The statistics, in plain English

A standardised mean difference near -0.9 for pain is a large effect, but the evidence is low certainty, so the true effect may be smaller. Rankings in network meta-analyses can shift with a few new trials.