Your knees hurt. Should you train through it or rest?

The short answer: usually neither — you adjust the load rather than choosing between full training and full rest. For most ordinary knee pain after 40, pain up to about 3 out of 10 during activity, settling by the next morning, is acceptable to work through. Above 5, or pain still elevated 24 hours later, is a signal to reduce load, not to stop moving. Complete rest is rarely the right answer and often makes things worse. The exceptions are specific, and they’re at the end of this article.

What changes with age

Cartilage thins and its repair capacity slows. By 50, a majority of people have some radiographic change in the knee — and here’s the part that matters: imaging findings and symptoms correlate poorly. Plenty of people with visible osteoarthritis on an X-ray have no pain at all, and plenty with severe pain have unremarkable scans. A scan describes the tissue. It does not tell you what the knee can do.

Meanwhile the quadriceps weaken, and they weaken fast when unused — measurable loss begins within a couple of weeks of reduced activity, and it’s quicker after 50 than at 30. Since the quadriceps absorb load that would otherwise pass through the joint, weakness and knee pain reinforce each other. That loop is the actual mechanism behind most slow-declining knees, and it’s the loop that rest feeds.

Tendons also change. They tolerate load fine but adapt more slowly to changes in load, which is why so much midlife knee pain traces to a step-up in activity three weeks earlier rather than to anything structural.

What the evidence supports

Exercise therapy has been shown to reduce pain and improve function in knee osteoarthritis. This is one of the better-evidenced interventions in musculoskeletal medicine — consistent across many trials, and recommended as first-line treatment by essentially every major clinical guideline including NICE and the NIH. The effect on pain is broadly comparable to what common painkillers achieve, with the difference that exercise also improves the function underneath.

Strengthening the quadriceps and hips has been shown to help specifically. Both isolated strengthening and general lower-limb programmes work. Notably, the type of exercise matters less than doing it consistently — trials comparing different modalities usually find similar results, which is quietly liberating. Do what your knee tolerates.

Cartilage responds to cyclical loading. It has no blood supply and depends on compression and release to move nutrients in and waste out. Research suggests moderate regular loading is protective rather than destructive, and the long-held fear that walking or running “wears out” knees isn’t supported — recreational runners have equal or lower rates of knee osteoarthritis than sedentary people.

Prolonged rest has been shown to worsen outcomes. Strength drops, joints stiffen, pain sensitivity increases, and confidence erodes. The old advice to rest a sore knee until it feels better creates a knee that feels worse for longer.

The traffic-light rule is a reasonable practical framework. It comes from tendon rehabilitation research and is widely applied to knee pain. Rate your pain 0–10 during and after the activity:

• Green (0–3): acceptable. Continue at this load.
• Amber (4–5): proceed with caution. Keep the session but reduce range, weight, or volume. Watch what the next 24 hours does.
• Red (6+): too much. Reduce the load this session — lighter, shallower, fewer reps — rather than cancelling it.

The second half of the rule is the more important one: pain should return to your normal baseline within 24 hours, and should not be worse the next morning. A knee that’s sore during a session but the same as always by breakfast has been loaded appropriately. A knee that’s stiff and swollen the next day was overloaded, regardless of how it felt at the time.

Weight matters mechanically and chemically. Each kilogram of body weight translates to several kilograms of force through the knee when walking, and research suggests fat tissue also contributes inflammatory signalling independent of the mechanical load. Modest weight loss combined with exercise outperforms either alone in trials.

What to do about it

Keep moving, change the variables. Nearly every painful knee tolerates something. If squatting to a chair hurts at 6, squat to a higher chair. If walking 40 minutes hurts, walk 20 twice. If depth is the problem, reduce range and load the range that doesn’t hurt. Cycling and swimming are useful substitutes when weight-bearing is the specific issue, though they don’t build the same leg strength.

Build the quadriceps deliberately. Sit-to-stands, step-ups to a low step, leg extensions in a pain-free range, wall sits. Two or three sessions a week. Expect this to take eight to twelve weeks to change how the knee feels — this is the slowest-acting and most durable of the interventions here.

Change one thing at a time. If pain flared after you started something new, the change is more likely the cause than your cartilage. Back the new thing off by about half and rebuild over three to four weeks.

Use the 24-hour check, not the in-session feeling. Write it down for two weeks if you’re unsure. Most people misjudge this by memory and are surprised by the pattern that emerges.

Expect flares, and don’t treat them as setbacks. A flare that settles in a few days is normal in an arthritic or previously injured knee. Reduce load for those days, then return — don’t reset to zero.

What we don’t know

How much of the pain in an arthritic knee comes from the joint itself versus from a nervous system that has become sensitised is genuinely unsettled, and it probably varies a great deal between people. That uncertainty is why two people with identical scans can need very different approaches.

We also don’t know the optimal dose. Trials show exercise works; they don’t converge on how much, how heavy, or how often, and the honest reading is that a wide range of doses helps. The traffic-light thresholds themselves are a pragmatic convention rather than a validated cut-off — they’re a decent default, not a law.

Whether specific loading protocols can meaningfully change cartilage structure, as opposed to improving symptoms and function, remains an open question. The symptomatic and functional benefits are solid. Structural claims are not.

Supplements marketed for joints — glucosamine and chondroitin most prominently — have been studied extensively with mixed and largely disappointing results. They may help some individuals; the trial evidence doesn’t support them as a general recommendation.

When to talk to a clinician

Some things override the traffic-light rule entirely. See someone promptly if you have:

• A hot, red, swollen knee with fever or feeling unwell — this needs urgent assessment the same day
• Swelling that appeared rapidly, within hours of an injury
• Locking — the knee catching and refusing to straighten
• Giving way — the knee buckling under you without warning
• Inability to bear weight, or an obvious deformity after an injury
• Pain that wakes you at night or is present at rest, unrelated to activity
• Numbness, tingling, or weakness below the knee
• Calf pain, warmth or swelling on one side — this is a circulation question, not a joint one

Beyond the red flags, it’s reasonable to get an assessment if pain has lasted more than six weeks without improving, if it’s limiting things you care about, or if you simply want someone to watch you move and tell you what’s driving it. A physiotherapist can identify a specific strength or movement pattern faster than trial and error will, and knowing which structure is irritable makes the traffic-light rule considerably easier to apply.