Short answer: Every major international guideline makes exercise a core treatment for knee osteoarthritis, and none recommends rest. But the 2024 Cochrane review of 139 trials found pain improvements of 8.7 to 13.1 points on a 100-point scale — at or below the threshold for a clinically meaningful change. The strongest case for training an arthritic knee is structural, not analgesic.
The X-ray comes back and there is a word on it. Somebody — a GP, a colleague, a forum — tells you that squatting is what did it, and that you should take it easy now.
Knees are not even the biggest problem lifters have. They came third and fourth in the two injury surveys behind our ranking of where lifters actually get hurt. But the knee is the one that arrives with a diagnosis attached, and a diagnosis changes how people behave.
Here is the problem with taking it easy: nobody has ever run a trial showing it works. There is no guideline anywhere that recommends rest for knee osteoarthritis. What there is instead is a large, messy, and more honest body of evidence than either side of this argument usually admits.
It says train. It does not say what most people think it says about how.
What the guidelines say, and how confident they actually are
Both the 2019 American College of Rheumatology guideline and the 2019 OARSI guideline list structured land-based exercise as a core treatment for knee osteoarthritis — the first thing you do, not the thing you try after injections fail. The ACR strongly recommends both aerobic and resistance exercise. Neither document recommends rest, activity avoidance, or protecting the joint from load.
Then you open the evidence those guidelines sit on.
The 2024 Cochrane review is the largest synthesis that exists: 139 trials, 12,468 participants, exercise programmes running anywhere from 2 to 104 weeks. It splits the question three ways depending on what exercise is compared against.
Now the number that matters. The threshold researchers use for a change a patient would actually notice — the minimal important difference — is 12 points for pain and 13 for physical function. Two of those three estimates sit below it. The review’s own conclusion is that the confidence intervals “either did not reach these thresholds or included both a clinically important and clinically unimportant improvement.”
This is a smaller claim than the one the fitness industry makes. It is worth sitting with rather than skipping past.
It is also a shrinking claim. The previous version of this same Cochrane review, published in 2015, pooled 44 trials and 3,537 participants and reported a 12-point pain reduction — right at the threshold. Nine years and 95 more trials later, the estimate came down. That is what usually happens when a field accumulates larger and better-controlled studies, and it is a reason to trust the newer number more, not less.
The trial that should change how lifters read this
If exercise helps, more load should help more. That is the assumption almost every lifter brings to this question.
It was tested directly. The START trial, published in JAMA in 2021, randomised 377 adults aged 55 and over with painful, radiographically confirmed knee osteoarthritis into three groups for 18 months.
High-intensity training did not significantly reduce knee pain compared with low-intensity training or the control group. It also did not significantly reduce knee joint compressive force.
Read that in both directions, because it cuts both ways.
Against the “load it heavy” position: eighteen months at 75–90% of 1RM bought no pain advantage over training at roughly a third of that. If you are grinding heavy sets because you believe the loading itself is therapeutic for the joint, this is the best available test of that belief and it came back negative.
Against the “protect the joint” position: eighteen months at 75–90% of 1RM did not make anything worse either. No excess pain, no increase in joint compressive force. The heavy group was not harmed.
That is the actual state of the evidence on load: heavy is not required, and heavy is not dangerous.
So why train at all
Because pain relief was never the strongest argument, and the structural argument is more robust than the analgesic one.
Strength predicts who gets worse. In the MOST cohort — 3,026 adults aged 50 to 79 followed prospectively — greater quadriceps strength was associated with lower risk of incident symptomatic knee osteoarthritis, and with less progression of joint space narrowing and cartilage loss in women. Weakness is not just a consequence of an arthritic knee. It runs ahead of it.
Training appears to slow structural progression. A 30-month randomised trial in 221 adults (mean age 69) compared lower-limb strength training against range-of-motion exercise. The strength training group preserved strength and showed less progression of joint space narrowing.
And at scale, in the real world, it works better than the trials suggest. The GLA:D registry — a nationwide exercise-and-education programme, 9,825 participants — reported pain improvements of 12.4 points at 3 months and 13.7 points at 12 months, above the 12-point threshold and holding at a year. Registry data is not a randomised trial and the comparison group is weaker, but it is the largest picture available of what happens when this is delivered properly and people keep doing it.
The comparison nobody offers you
The alternative on the table is not “no risk.” It is a different risk that nobody quantifies when they tell you to take it easy.
The closest clean data comes from running rather than lifting, so read it as an analogy rather than a direct transfer.
A 2017 systematic review reported hip or knee osteoarthritis prevalence of 3.5% in recreational runners, 10.2% in sedentary controls, and 13.3% in competitive and elite runners.
The shape of that curve is the point. Doing nothing was worse than doing a moderate amount. Doing a professional amount was worse again. The risk was never a straight line where less is always safer — and “rest” sits on the wrong end of it.
What to do with this
1. Stop looking for the pain fix and start protecting the joint. The pain effect of exercise is real and modest — somewhere around 10 points on a 100-point scale. The structural case is stronger. Train the knee because strength tracks with slower progression, not because you expect the ache to disappear.
2. Stop assuming heavier is more therapeutic. Eighteen months at 75–90% of 1RM beat nothing and beat 30–40% of 1RM by nothing. If heavy squatting is what you want to keep doing, the data says you can. It does not say it is treatment.
3. Judge it at twelve months, not twelve weeks. The trial effects are measured immediately post-programme and fade after it stops. The registry data holds at a year because people stay in it. Whatever you choose has to be something you will still be doing next August. The same caution applies to the broader assumption that a body past 35 needs protecting — the recovery data does not support it.
Frequently asked questions
Can you still lift weights with knee osteoarthritis?
The evidence supports it. In an 18-month randomised trial of 377 adults with knee osteoarthritis, strength training at 75–90% of 1RM produced no increase in knee pain and no increase in knee joint compressive force compared with much lighter training or a control group. Every major guideline lists exercise as a core treatment and none recommends rest.
Does squatting make arthritis worse?
Not according to the available trial data. The largest long-term trial of heavy strength training in people with knee osteoarthritis found no increase in pain or joint compressive force over 18 months. A separate 30-month trial found that strength training slowed the progression of joint space narrowing compared with range-of-motion exercise.
How much does exercise actually reduce arthritis pain?
Less than commonly claimed. The 2024 Cochrane review of 139 trials and 12,468 participants found pain improvements between 8.7 and 13.1 points on a 0–100 scale depending on the comparison group. The threshold for a change a patient would notice is around 12 points, so several of those estimates fall at or below it.
Is heavy lifting better than light lifting for arthritic knees?
The direct comparison says no. The START trial randomised adults with knee osteoarthritis to high-intensity training at 75–90% of 1RM or low-intensity training at 30–40% of 1RM for 18 months and found no significant difference in knee pain or joint compressive force between them.
Is resting better for arthritic knees?
No guideline recommends it and no trial supports it. In a systematic review of running and osteoarthritis, sedentary controls had a hip or knee osteoarthritis prevalence of 10.2% compared with 3.5% in recreational runners. Prospective cohort data also links lower quadriceps strength to higher risk of symptomatic knee osteoarthritis.
Should I keep training if my knee hurts during a set?
That is a question for a clinician who can examine the knee, and the answer depends on findings an article cannot see. What the research can say is that structured exercise programmes for knee osteoarthritis do not appear to increase withdrawal rates, and that adverse events, while uncommon, were reported more often in exercise groups than usual-care groups in the 2024 Cochrane review.
How we sourced this
Every figure above comes from a published systematic review, randomised trial, prospective cohort, or national registry. Where estimates disagree, all of them are shown. This article is not medical advice and does not replace an assessment from a clinician who can examine the actual knee.
- Lawford BJ, et al. Exercise for osteoarthritis of the knee. Cochrane Database of Systematic Reviews, 2024 (CD004376.pub4). cochranelibrary.com
- Fransen M, et al. Exercise for osteoarthritis of the knee. Cochrane Database of Systematic Reviews, 2015. pubmed.ncbi.nlm.nih.gov/26405113
- Messier SP, et al. Effect of High-Intensity Strength Training on Knee Pain and Knee Joint Compressive Forces Among Adults With Knee Osteoarthritis: The START Randomized Clinical Trial. JAMA, 2021. pubmed.ncbi.nlm.nih.gov/33591346
- Mikesky AE, et al. Effects of strength training on the incidence and progression of knee osteoarthritis. Arthritis & Rheumatism, 2006. pubmed.ncbi.nlm.nih.gov/17013851
- Segal NA, Glass NA, et al. Is quadriceps muscle weakness a risk factor for incident or progressive knee osteoarthritis? (MOST cohort), 2012. pubmed.ncbi.nlm.nih.gov/22293767
- Evidence That Good Life With Osteoarthritis in Denmark (GLA:D), an Exercise and Education Program, Is Beneficial for Osteoarthritis. The Journal of Rheumatology, 2024 — registry analysis of 9,825 participants. jrheum.org
- Kolasinski SL, et al. 2019 ACR/Arthritis Foundation Guideline for the Management of Osteoarthritis of the Hand, Hip, and Knee. pubmed.ncbi.nlm.nih.gov/31908149
- Alentorn-Geli E, et al. The Association of Recreational and Competitive Running With Hip and Knee Osteoarthritis: A Systematic Review and Meta-analysis. JOSPT, 2017. jospt.org
IRONBACKPACK’s evidence desk for people who have been training long enough to have a list. No programs to sell, no supplements, no motivation. Just what the research says and what it means on the gym floor.
Get the next one by email. One post a week. Unsubscribe whenever.
SUBSCRIBE