Is Running Bad for Your Knees? What 50 Years of Research Shows

Running ruins your knees. It is the most durable piece of gym-floor folklore there is, and the public holds it harder than clinicians do. A survey of 4,521 people across seven languages found 68% of the general public called regular running healthy for the knees, against 86% of health care professionals, and 60% of the public believed hard surfaces raise arthritis risk compared with 36% of clinicians (multilanguage survey, 2022). The evidence sits with the clinicians. Here is what the studies show, and what does put a knee at risk.

What five decades of studies found

The largest pooled analysis on this question combined 17 studies covering 114,829 people. Hip or knee osteoarthritis showed up in 3.5% of recreational runners. In non-running controls, 10.2%. In competitive runners, 13.3% (Alentorn-Geli et al., 2017). Recreational running was associated with roughly a third of the arthritis prevalence seen in people who did not run.

The cohort data repeats it. A cross-sectional study of 2,637 adults from the Osteoarthritis Initiative found symptomatic knee osteoarthritis in 23.5% of people with any history of running versus 29.6% of non-runners, and frequent knee pain in 35.1% versus 41.1% (Lo et al., 2017). After adjusting for age, sex and BMI, running history was never associated with higher risk.

A 2023 systematic review pooled 17 studies and 14,141 participants, 7,194 of them runners, and found no difference in cartilage thickness on MRI between runners and non-runners, plus knee pain in 28.2% of runners against 41% of non-runners (Dhillon et al., 2023). A broader 2024 overview of the evidence reported that recreational running up to 25 miles a week showed no link to structural knee deterioration. It also reported 54% lower odds of knee surgery in runners than non-runners (evidence overview, 2024).

None of this proves running protects cartilage. It does mean the wear-and-tear model most people carry around has no support in the data.

High impact does not mean high damage

Around 8 times bodyweight. That is the peak knee joint contact force during running, against roughly 2.7 times bodyweight when walking, and it is where the wear-and-tear intuition comes from. If joints were bearings, that would settle it.

They are not. Measure the load per unit of distance covered rather than the load per step, and running and walking come out nearly identical: 0.80 versus 0.75 bodyweights per meter (Miller et al., 2014). A runner spends far less time in contact with the ground, and covers far more distance per stride. The peak is higher. The dose per kilometer is the same.

Cartilage also responds to that dose rather than accumulating damage from it. Eight healthy male runners were scanned before running, immediately after, and 24 hours later. T1rho relaxation times, a proxy for proteoglycan content, dropped 4% after a 3-mile run and 11% after a 10-mile run, and at 24 hours neither distance showed a difference from pre-exercise values (a T1rho MRI study). Cartilage compresses under load, sheds water, rehydrates. That is what living tissue does.

One more point sinks the bearing analogy. Cartilage has no blood supply, so it depends on compression and release to move nutrients in and waste out. Sitting still does not spare it. It starves it. That may be part of why sedentary controls keep showing up with more arthritis than the runners in these datasets, and why the medical advice to stop running rarely comes with an equally specific replacement.

Recreational and competitive are different exposures

The meta-analysis put competitive runners at 13.3%, higher than sedentary controls, and that number gets quoted as proof running is dangerous. It deserves an honest reading rather than a dismissal.

Competitive in those studies meant professional or elite athletes, or people racing internationally. Often decades of it, at volumes most people never touch, with the acute injuries and surgeries a career like that collects along the way. Different dose. Different injury history.

The recreational picture is clearer. A survey of 3,804 marathon runners with a mean age of about 44 found hip or knee arthritis in 7.3% of them, and no association between arthritis and years spent running, number of marathons completed, weekly mileage, or pace (Hartwell et al., 2023). Age, BMI, family history, and previous injury or surgery did predict it, and around a quarter of those runners had been told by a physician to cut back or stop.

What does put a knee at risk

Four risk factors carry most of the weight. Running is not among them.

• A previous knee injury. The big one. Pooled data put the odds of knee osteoarthritis at roughly four times higher after an ACL tear, and around six times higher after a meniscal injury or a combined injury (a systematic review and meta analysis); a review of post-traumatic risk factors found ACL reconstruction combined with total medial meniscectomy raised the odds of osteoarthritis more than threefold against isolated reconstruction (the OPTIKNEE systematic review). Avoiding the injury in the first place is the highest-value knee intervention there is.
• Bodyweight. One kilogram lost cut compressive knee force during walking by about 40 newtons per step, in 142 overweight older adults with knee osteoarthritis (Messier et al., 2005). Multiply that across thousands of steps a day and it adds up faster than any change you can make to your stride.
• Muscle. In 2,492 adults followed for four years, sarcopenia was associated with more than double the odds of developing symptomatic knee osteoarthritis, at OR 2.29 (Osteoarthritis Initiative cohort data). Weak legs mean less of the load gets absorbed by muscle before it reaches the joint.
• Age and family history. Neither is negotiable. Both showed up as predictors in the marathon survey.

Three of those four respond to training. Running does not appear on the list. Quit running to protect your knees and you lose muscle mass and make weight control harder, which moves you toward the risk factors that do matter. For how load tolerance is built rather than assumed, see tissue capacity for runners.

The single session is the risk, not the week

The most useful running-injury finding of the last decade landed in 2025. The Garmin-RUNSAFE study followed 5,205 runners with a mean age of 45.8 across 18 months and 588,071 recorded sessions. A total of 1,820 of them, 35%, reported a running-related injury, and 1,311 of those were overuse (Schuster Brandt Frandsen et al., 2025).

The week did not predict injury. The individual run did. Compared with sessions no more than 10% longer than the runner's longest run of the previous 30 days, overuse injury rates rose with the size of the spike:

• 10-30% longer than the longest run of the past 30 days: hazard rate ratio 1.64 (95% CI 1.31 to 2.05)
• 30-100% longer: HRR 1.52 (95% CI 1.16 to 2.00)
• More than 100% longer: HRR 2.28 (95% CI 1.50 to 3.48)

In that data the acute:chronic workload ratio showed a negative dose-response, meaning the bigger weekly spikes came with lower injury rates, and the week-to-week ratio showed no relationship at all. The load metrics built into watches and coaching software pointed the wrong way or nowhere.

The practical version is short. Look at your longest run in the last 30 days, keep today's run inside about 10% of it, and when you want a longer long run, move the ceiling up in steps rather than in one jump. The ceiling resets downward every time you take a few weeks off. That is also the rule that protects a comeback, which returning to running after injury covers in phases.

If your knees already hurt

An aching knee is a reason to change how you run. It is not always a reason to stop. Among Osteoarthritis Initiative participants aged 50 and over with existing knee osteoarthritis, self-selected runners showed no worsening of pain and no faster structural progression over 48 months than non-runners (Lo et al., 2018). Get a diagnosis first. Then work the three levers that have evidence behind them.

• Raise your cadence. In 30 healthy runners, increasing step rate to 110% of preferred cut peak patellofemoral joint force by 14% and stance-phase impulse by 20% (a joint-force modelling study). Same distance, smaller and quicker steps, less force per contact. A cadence calculator to find your target step rate gives you the number to run to, and a running form analyzer that reads a phone video of your stride shows what your current one looks like.
• Lift. A meta-analysis of 25 trials and 26,610 participants found strength training cut sports injuries to under a third of the control rate (RR 0.315), and exercise programmes cut overuse injuries roughly in half (RR 0.527) (Lauersen et al., 2014). Squats, hinges, single-leg work, heavy calf raises. Twice a week. Strength training for runners has the programming.
• Progress the load slowly. Cartilage, tendon and bone adapt over months, not weeks. Run-walk intervals let you accumulate volume while keeping any single session inside the tissue's current tolerance: four sets of 5 minutes running with 2 minutes walking gives you 20 minutes of running load in a 28-minute session, and puts a ceiling on how long any one bout of impact lasts.

Sharp pain that changes your gait, swelling that lasts overnight, pain that gets worse mile by mile: stop and get it looked at. Stiffness that eases in the first ten minutes is the ordinary arthritic knee. It usually feels better on the days you move.

Running after 50

From 1984 to 2002, the longest prospective look at this followed 45 distance runners and 53 controls with knee radiographs. At the end, runners did not have more radiographic osteoarthritis than controls, 20% versus 32%, a difference that did not reach significance. What predicted the worst outcomes was higher initial BMI and existing joint damage (Chakravarty et al., 2008). Eighteen years of mileage in older adults did not accelerate the radiographs.

What changes with age is the margin for error. Sarcopenia risk climbs, so the strength work moves from optional to load-bearing. Recovery between hard sessions stretches. Two quality days a week is usually enough. And a layoff costs you more at 55 than it did at 30, so the ceiling you come back to sits lower, which makes the single-session rule more valuable rather than less. The RUNSAFE cohort itself had a mean age of 45.8, so those hazard ratios describe masters runners rather than college athletes. Masters athlete training after 40 covers what else shifts and what holds.

Bodyweight is the quiet variable here. Most people gain slowly through their forties and fifties, and every kilogram travels with them into every step they take. Shoes, insoles and running surfaces get more attention than they have earned. Holding your weight steady is the bigger lever.

How Movement Rebels fits

Every knee-protective rule here runs on knowing your own recent history. That is the thing a runner forgets between Tuesday and Sunday. The Movement Rebels coach reads your Garmin and Apple Health syncs, so it sees your last 14 sessions, with distance, pace, cadence and per-kilometer splits on the runs, what sits on your calendar for the week ahead, the lifts you logged, and your morning check-in, including the soreness score you gave your legs today.

That combination is what makes the plan honest. Three mornings of soreness at 7 on your check-in, alongside a cadence that has drifted down across your recent long runs, is a different week than the calendar assumed. The coach can move a session, rebuild the week, or hold the calendar in place while you sort the knee out, then put the strength work back in once you are loading again.

Look at your last 30 days of runs. The longest one is your ceiling. Everything else on this list is easier once you know that number.