The short answer
- Across 25 studies covering 125,810 people, hip and knee arthritis was present in 3.5% of recreational runners, 10.2% of non-running controls, and 13.3% of competitive runners.
- Recreational running sat lower than doing nothing. Elite and professional-level running sat higher.
- In people over 50 who already had knee arthritis, taking up or continuing running over four years did not worsen their pain or their x-rays.
- In that group, runners were more likely than non-runners to have their frequent knee pain resolve.
- None of this proves running protects joints, because runners differ from non-runners in other ways. It does mean the wear story is not supported.
Everyone who runs has been told that their knees will pay for it eventually. It is said kindly, usually by someone who does not run. It is the single most durable myth in this field, and the evidence against it is unusually large.
The comparison that matters
Read the three prevalence figures in order. Recreational runners: 3.5%. People who did not run: 10.2%. The group doing nothing had arthritis roughly three times as often as the group running recreationally.
Competitive runners, meaning professional and international-level athletes, sat highest at 13.3%. So the relationship is not a straight line where more is always better. It is closer to a U, where a moderate amount sits at the bottom.
What if you already have it
A separate and more practical question, and it has been studied directly.
Four years of running in knees that already had arthritis, and the x-rays did not get worse. The pain did not get worse. The people who ran were somewhat more likely to have their pain go away.
The authors are careful about why, and so are we. This was self-selected running: people chose how much to do, and that choice was probably influenced by how their knee felt. That is a real limitation, and it is also close to how you would actually do it.
Why the x-ray is not the whole story
Arthritis on imaging and arthritis as an experience are loosely coupled at best.
Plenty of people walk around with changes on a scan and no symptoms. Plenty of people have significant pain with modest changes. That is why the plan is built around what your knee does rather than what the report says.
What actually helps
The thing that most reliably improves an arthritic knee is the same thing that most reliably prevents running injuries.
Strength training has the strongest evidence of any intervention for reducing injury in the wider sports literature4, and it is also first-line treatment for knee osteoarthritis in essentially every guideline. Quadriceps and hip strength, built over months.
Alongside that: keep the load in a range your knee tolerates, use the same pain rule you would use for anything else, and expect good weeks and bad weeks rather than a smooth line.
What does not help is stopping. Deconditioning takes strength away from the joint that needs it, and a knee that has not been loaded in a year tolerates less than the one that has.
Where the caution is real
Two things in the evidence above deserve to be said plainly rather than glossed over.
Competitive, elite-level running was associated with higher arthritis prevalence than controls. That is a different exposure from recreational running, in volume and in years, and often with a history of significant injury behind it.
And previous joint injury is the factor the authors specifically flag as a possible confounder. A knee that has had a serious injury is at higher risk regardless of what you do afterwards, which is an argument for treating injuries properly rather than for avoiding running.
When to get it looked at
The last one is the reason many people arrive here. It is a legitimate question and worth asking properly.