Does running ruin your knees?

A systematic review and meta-analysis in the Journal of Orthopaedic & Sports Physical Therapy included 25 studies of 125,810 people and pooled 17 of them, covering 114,829 people. The prevalence of hip and knee osteoarthritis came out at 3.5% among recreational runners (95% CI 3.4% to 3.6%), 10.2% among non-running controls (95% CI 9.9% to 10.6%) and 13.3% among competitive runners (95% CI 11.6% to 15.2%). Those three percentages are the ones that travel around the internet, and on their own they look like a knockout. They are not, and the same paper says so in the next sentence: when the review pooled the comparison study by study as an odds ratio, recreational runners came in at 0.86 (95% CI 0.69 to 1.07) and competitive runners at 1.34 (95% CI 0.97 to 1.86), both against non-running controls. Both of those ranges include 1 — the value that means no difference. So neither runner group was significantly different from the controls. What was significant was the gap between the two runner groups themselves (P<.001). Narrower cuts of the same data do sit below 1 for recreational runners: 0.83 for the knee alone (0.70 to 0.99), 0.78 for men (0.68 to 0.89) and 0.54 for women (0.41 to 0.71). They rest on fewer studies, and the men's and women's figures lean heavily on one study, so they do not replace the overall result.

Alentorn-Geli E et al. 2017

A pooled prevalence adds up cases and people across studies. A pooled odds ratio compares runners with non-runners inside each study first, then combines those comparisons. The second is the fairer question, because runner groups and control groups in different studies differ wildly in age, sex, country and decade, and simply stacking their percentages side by side compares populations rather than people. The narrowness of the range around 3.5% is a symptom of this: it reflects how many people were counted, not how well the groups match. One study supplies about 89% of the recreational runners, compared with walkers older than they were, so the 3.5% mostly reflects that study. So the honest headline from this review is a null. That is a genuinely useful result — the claim under examination is that running wrecks knees, and the largest pooled analysis of the question does not find it — but it is an absence of harm, not a demonstration of benefit, and those are different sentences.

Alentorn-Geli E et al. 2017

The recreational-versus-competitive contrast is the honest heart of this topic, and it does not run in one direction. Competitive runners — defined in the review as professional or elite athletes, or people competing internationally — had the highest prevalence of the three groups at 13.3%, higher than the 10.2% in controls. Their odds ratio of 1.34 was not statistically significant on its own, but the difference between the competitive and recreational groups was. The review also split runners by years. Running for less than 15 years was associated with lower odds than in controls (0.6, 95% CI 0.49 to 0.73). Running for more than 15 years came out at 1.01 (95% CI 0.72 to 1.43) from 3 studies: no difference from controls, though significantly different from the shorter group (P=.01). The catch is that every under-15-year study was of recreational runners and every over-15-year study was of competitive runners, so those may be one split rather than two, and the review has no information on recreational running beyond 15 years. Its authors conclude that a more sedentary lifestyle and long exposure to high-volume or high-intensity running are both associated with hip and knee osteoarthritis, and also that no clear conclusion could be drawn beyond 15 years. So there are signs that the elite group differs, not a measured curve with a rising end. None of this sets a safe limit: the review itself says the amount of running that is safe for the joints could not be determined. There is also no number anywhere in this fix, on purpose: nothing in this evidence set tests a progression rate, a weekly distance target or a safe-increase percentage, so any figure you have heard for how fast to build up did not come from here.

Alentorn-Geli E et al. 2017

Every study in this evidence set is observational, and the specific weakness here is worse than usual. Nobody is assigned to be a runner. People choose it, and — this is the crucial part — people whose knees have started to hurt stop. That means any group labelled "runners" has been quietly filtered for having knees that still work, and any group labelled "non-runners" has quietly absorbed some of the people whose joints already gave them trouble. That single mechanism could produce the entire pattern above without running having done anything at all. Runners also tend to be lighter, and body weight is one of the strongest known risk factors for knee osteoarthritis. The review's own authors put it plainly: it was not possible to determine whether these associations were causal, or confounded by other risk factors such as a previous injury. Take that seriously in both directions. It undercuts any claim that running is good for joints just as firmly as it undercuts the claim that running destroys them.

Alentorn-Geli E et al. 2017

A 2023 review pooled 9 case-control studies covering 12,273 participants, of whom 1,272 were runners, and asked specifically whether weekly distance mattered. It found no difference in knee osteoarthritis prevalence between runners and a control group running under 8 km a week (0.97, 95% CI 0.56 to 1.68), and no higher prevalence in any distance band: 8 to 32.1 km a week gave 1.17 (0.77 to 1.80), 32.2 to 48 km gave 1.04 (0.48 to 2.31), and over 48 km gave 0.62 (0.35 to 1.10). Every one of those ranges crosses 1, and the reviewers rated most of the included studies as high or very high risk of bias, so this is an absence of evidence rather than evidence of absence. Separately, a 2017 review in the American Journal of Sports Medicine examined 15 studies and concluded flatly that with this evidence it is not possible to determine the role of running in knee osteoarthritis — its included studies suggested no association with a diagnosis, a positive association with a diagnosis, and a lower rate of knee surgery in runners, all at once. Three independent teams, three different methods, and the same destination: the feared association has not been demonstrated, and neither has its opposite.

Burfield M et al. 2023 · Timmins KA et al. 2017

Everything above describes populations of people with, for the most part, working knees. It is not guidance for a knee that hurts today. If you already have knee or hip pain, a diagnosed joint condition, a previous serious knee injury such as a torn cruciate ligament or meniscus, or a joint replacement, none of this evidence speaks to your situation, and what you should do about it is a conversation with a doctor or physiotherapist rather than something to read off a chart of prevalences. A previous joint injury is one of the strongest known risk factors for osteoarthritis, and it is the confounder the anchor review names explicitly. This module exists to remove a false reason not to start, for people who had no other reason. It is not a reason to override pain.

Alentorn-Geli E et al. 2017

The best pooled evidence does not show that it does. In a review that pooled 17 studies of 114,829 people, recreational runners had odds of hip and/or knee osteoarthritis of 0.86 compared with non-runners (95% CI 0.69 to 1.07), and competitive runners 1.34 (95% CI 0.97 to 1.86) — neither significantly different from them. A separate meta-analysis of 12,273 participants found no difference either. Everything here is observational, and people whose knees hurt stop running, so none of it can prove running is harmless. But the specific fear has been looked for repeatedly by independent teams and has not been found.

This evidence does not answer that, and we are not going to pretend it does. Every figure here describes groups of people who were, for the most part, running or not running without a current problem, and a knee that hurts today is a different question with a different answer. Take it to a doctor or a physiotherapist. The point of this module is narrower and worth keeping separate: if your knees are fine and the only thing stopping you is the belief that running would wreck them, that belief is not supported by the pooled evidence.

Nothing in this evidence set examines footwear, orthotics or running surface, so we have no figure to offer and will not invent one. What the reviews above compared was running against not running, and running at different weekly distances — not one kind of shoe against another. The fix here deliberately requires no purchase of any kind: it is a jog between two lampposts on a walk you were already taking, in whatever you already own.