Why Your Knees Hurt at 35

It is usually not the cartilage

The cartilage theory has an obvious appeal: it explains an ache with a picture. Get the knee scanned in your mid-30s and there is a reasonable chance something turns up. The problem is what turns up in people who feel perfectly fine.

A 2019 systematic review pooled 63 studies covering 5,397 knees in adults with no symptoms and no history of injury. Cartilage defects were present in 24 percent of them. Meniscal tears were present in 10 percent. The numbers climb with age, from roughly 4 to 14 percent in adults under 40 to 19 to 43 percent in adults over 40. These are knees that do not hurt. The authors' own conclusion is that such findings should be read in the context of the clinical picture rather than treated as the explanation on their own.

So a scan showing wear in a 35-year-old knee has told you something roughly as specific as a photograph showing grey hair. It is real, it is age-related, and on its own it does not establish why this knee is sore now when the identical finding sits silently in a quarter of the people around you.

Where the hip story breaks

The upstream account is more sophisticated and much better liked. The hip stabilisers, the gluteus medius in particular, control how the femur behaves under load. Let them weaken and the thigh drifts inward, the kneecap tracks badly, and the knee pays for a problem it did not create. It is mechanically plausible, it explains the pain without blaming the joint, and it comes with a fixable culprit.

It has also been tested the only way that can settle it: measure people before they hurt, then see who goes on to develop pain. Do that, and the story does not hold.

A 2014 meta-analysis separated the cross-sectional studies from the prospective ones and found they disagreed. In cross-sectional comparisons, adults with patellofemoral pain did have lower hip strength than pain-free people. In prospective studies, there was moderate-to-strong evidence of no association between isometric hip strength and the risk of developing it. The authors drew the conclusion that follows: reduced hip strength may be a result of the pain rather than its cause.

That single distinction dissolves most of the folklore. If you measure people who already hurt, you find weak hips, because hurting for months makes people move less and load that limb less. Read backwards, it looks like a cause. Measured forwards, it is a consequence.

A 2019 review pooled 18 studies and 4,818 participants, of whom 483 went on to develop patellofemoral pain, and reached the same place: hip weakness was not a risk factor. In adolescents it found the opposite of the folklore, with greater hip abduction strength associated with higher risk. And there is a paper in the American Journal of Sports Medicine, in young female athletes, titled plainly enough to end the argument on its own: hip strength is greater in athletes who subsequently develop patellofemoral pain.

What the evidence does predict

The same 2019 review is more useful for what it ruled out than for what it found. Age, height, weight, body mass index, body fat and Q angle were all reported as not being risk factors for future patellofemoral pain. That is a long list of things people are routinely told to worry about, cleared in one analysis.

One modifiable factor did survive, and it is not the hip. Quadriceps weakness predicted future pain in military recruits, particularly once normalised for body mass. Worth stating the scope precisely: that finding comes from military populations entering a sharp, uniform jump in training load, and it does not automatically transfer to a 35-year-old adding a weekly run. But it is the one strength variable that predicted anything, and it sits at the knee rather than above it.

1. Not a risk factor

   Hip weakness

   Multiple pooled analyses, effect sizes clustered near zero. In adolescents, greater hip abduction strength went the other way.

2. Not a risk factor

   Age, height, weight, BMI, body fat, Q angle

   Strong to moderate evidence against each, in the same pooled review.

3. A risk factor, in one population

   Quadriceps weakness

   Moderate evidence in military recruits, strongest when normalised to body mass.

4. Not what most people assume

   Running itself

   Recreational runners show less hip and knee osteoarthritis than non-runners, though the authors caution the association may be confounded.

That last line deserves its own moment, because the mid-30s is exactly when people quietly retire from running to protect their knees. A 2017 systematic review put hip and knee osteoarthritis at 3.5 percent in recreational runners, 10.2 percent in non-running controls, and 13.3 percent in competitive runners. The shape is a curve, not a slope: recreational running looks better than sitting still, and high-volume competitive running looks worse than both.

The authors are careful, and so is this article: they could not establish whether the association is causal or confounded by other factors, including previous injury. What the data will not support is the thing people actually do with it, which is to stop running in their thirties because the joint has a mileage limit.

And if it really is arthritis

Some readers will have a diagnosis rather than a suspicion, and the fear underneath the question is usually the same one: that loading a worn joint spends something that cannot be replaced. It is worth answering directly, because it is the belief that most often stops people moving.

A 2019 review gathered the trials that imaged cartilage before and after exercise programmes in people either at raised risk of knee osteoarthritis or already living with it. Across studies reporting cartilage thickness, volume and defects, the pattern was an absence of harm. The authors' conclusion is that knee joint loading exercise seems not to be harmful for articular cartilage in these groups, with the honest qualification that the quality of the underlying evidence was low.

On whether it helps, the evidence is considerably stronger. A Cochrane review of exercise for knee osteoarthritis found high-quality evidence that it reduces pain and moderate-quality evidence that it improves physical function. The effect is moderate immediately and smaller at two to six months, which sounds unimpressive until you read the comparison the authors draw: it is comparable with the estimates reported for non-steroidal anti-inflammatory drugs. They add a line reviews rarely include, that further research in this area is unlikely to change their findings.

So the exercise that a worn knee is often protected from performs about as well as the drug people take instead, and the imaging studies do not show it costing the joint anything. That is a different conversation from managing a fixed budget of cartilage.

Why the work still helps

Here is the part that makes this worth reading rather than merely debunking. The exercises were never the problem. Only the story attached to them was.

A Cochrane review found that exercise therapy for patellofemoral pain reduces pain and improves function, with some low-quality evidence that hip plus knee exercise beats knee exercise alone. A separate meta-analysis of proximal work concluded that a robust body of evidence supports including it in conservative management, and that combining proximal with quadriceps rehabilitation produced greater pain reduction and better function at one year. The 2018 international consensus statement recommends exercise therapy, and specifically does not recommend joint mobilisations in isolation or electrophysical agents.

So hip and quadriceps training is the best-supported thing you can do about a sore knee. It simply is not correcting the deficit it is advertised as correcting. It changes how much load the limb tolerates and how confidently it is used, which is a different claim from repairing a weakness that caused the problem.

The practical difference is real. If you believe a weak gluteus medius caused this, then a plateau means more clamshells, and a knee that still hurts after eight weeks of band walks means you have not fixed the cause yet. If you understand the work as building tolerance rather than correcting a defect, a plateau means the load is wrong, and the answer is usually to progress into heavier, slower, more knee-dominant strength work rather than to keep drilling the hip harder.

It also changes what you stop doing. Nothing in this literature supports retiring from running, chasing a scan, or treating an incidental cartilage finding as a verdict. The evidence supports loading the leg, progressing it patiently, and giving it longer than feels reasonable.

The knees this article is not about

Everything above is about an ache that arrives gradually on stairs. Two other knees look nothing like that and are emergencies rather than reasons to start a programme, and an article that spends its length telling you not to chase a scan owes you both of them plainly.

A hot, swollen, red knee, especially with a fever

The NHS lists the symptoms of septic arthritis as sudden severe pain in a joint, not being able to move or put weight on it, hot, swollen or red skin over it, and a high temperature, adding that not everyone has one. It says to ask for an urgent GP appointment or get help from NHS 111, and that symptoms usually develop quickly over a few days. The reason for that urgency is a clock rather than a queue: in a 2021 review for American Family Physician, subcartilaginous bone loss, cartilage destruction and permanent joint dysfunction can occur if appropriate antibiotic therapy is not started within 24 to 48 hours of onset. This is the one knee where waiting to see how it settles is the wrong instinct, and where the absence of a fever is not reassurance.

A painful, swollen calf

A deep vein thrombosis presents around the knee rather than in it, and it is easy to file as a strained calf. The NHS lists throbbing pain in one leg, usually in the calf or thigh, swelling in that leg, red, blue or darkened skin around the painful area, and swollen veins that are hard or sore to touch. It says to get urgent advice the same day if you think you have one, and to call 999 or go to A&E if those symptoms come with breathlessness or chest pain, because a clot that travels to the lungs is a pulmonary embolism and needs treatment straight away.

None of this is medical advice, and none of it replaces assessment by a clinician who can examine you. Sudden swelling, a knee that locks or gives way, or pain following a specific injury are also reasons to be seen rather than to start a programme. The gradual ache on stairs in your mid-30s is the one this evidence is about.