BackOctober 26, 202615 min readEmmaCentury

Runner's Knee: Why Your Knees Hurt When You Run (and How to Keep Training)

Patellofemoral pain is the most common reason runners end up in a clinic, and most of the advice online is wrong. Here's what the clinical guidelines actually recommend, which exercises matter, and how to keep running without making it worse.

Updated 2026-10-26Editorial standards
Runner's Knee: Why Your Knees Hurt When You Run (and How to Keep Training)

The pain doesn't usually arrive dramatically. There's no pop, no swelling, no moment you can point to. It just starts showing up on the downhill of your easy run, or when you stand up after sitting at your desk for an hour, or on the stairs at home.

That's the signature of patellofemoral pain, which almost everyone calls runner's knee. It's the most common injury seen in sports medicine clinics, and it has a reputation for hanging around. The 2016 international consensus on the condition described it plainly as a problem that tends to persist for years in a meaningful share of people.

Here's the frustrating part: most of what gets passed around about runner's knee is either outdated or backwards. Ice, rest until it stops, stretch your IT band, strengthen your inner quad, buy a special brace, stop running. Some of that is harmless. Some of it actively delays your recovery, because it keeps you from doing the one thing that has the best evidence behind it.

This is what the clinical guidelines actually say, what the evidence behind them looks like, and how to keep running while you fix it.

TL;DR: Runner's knee is patellofemoral pain: pain around or behind the kneecap that gets worse with loaded bending of the knee. It's not damage, and imaging usually won't explain it. The strongest treatment by a wide margin is progressive strength work combining hip and knee exercises, done consistently for at least six to twelve weeks. It's fine to keep running with pain you can tolerate, as long as you're not making it worse week to week. Gait changes help some runners: a 5 to 10% cadence increase, softer landings, and less hip collapse. Passive stuff like ultrasound, dry needling and manipulation doesn't work. Most people improve, but it takes months, not days, and the rehab is boring on purpose.

YouTube: Related video

What runner's knee actually is

The 2016 consensus statement defined the core criterion: pain around or behind the kneecap, aggravated by at least one activity that loads the joint while your knee is bent and bearing weight. Squatting, stairs, running, hopping, kneeling, sitting for a long time with the knee flexed. That's the whole diagnostic picture, and it's deliberately simple, because there's no scan or test that confirms it.

That last point matters more than people expect. Magnetic resonance imaging is not useful for diagnosing patellofemoral pain. The guidelines are explicit about this. Plenty of people with no knee pain have cartilage changes on MRI, and plenty of people with significant pain have a completely unremarkable scan. If you've had imaging that showed "mild chondral changes" and you've been told that's your problem, that's not a diagnosis, it's an incidental finding.

The old term "chondromalacia patellae" is worth retiring too. It describes softening of the cartilage on the back of the kneecap, and the consensus work moved away from it as a label for this pain, because the cartilage state and the pain experience don't line up.

What you're dealing with is a load-tolerance problem, not a structural defect. The joint is being asked to handle more than it's currently conditioned for, or the same amount with less capacity to absorb it, and the result is sensitised tissue around the kneecap.

Why it usually starts

The guidelines describe the mechanism as overuse: too much load magnitude, too much load frequency, or too fast a rate of increase. In practice that means a mileage jump, a new hill route, adding speed work to a base that wasn't ready, switching from treadmill to concrete, or a sudden return after a break.

There's a nice piece of supporting evidence for the load story. Runners with patellofemoral pain have shown increased patellar bone water content compared with pain-free controls, which suggests swelling in the bone under the kneecap. Another study found elevated bone metabolic activity in the joint, with a moderate correlation between tracer uptake and pain intensity. Something is being loaded beyond what the tissue is adapted to.

What the guidelines also dismantled is the folklore about body structure. A meta-analysis of two studies found that low isometric knee extension strength predicted who would develop patellofemoral pain. But the Q angle, static knee valgus, and measured dynamic knee valgus did not predict it. So the idea that your knees "track badly" because of your hip width or your alignment is not supported. Strength is a better predictor than shape.

What the guidelines actually recommend

The 2019 clinical practice guidelines from the Academy of Orthopaedic Physical Therapy graded interventions by strength of evidence. Here's the short version.

Intervention Grade What it means in practice
Exercise therapy, combined hip + knee A (should) The core treatment. Non-negotiable.
Hip-targeted exercise first, early on B (may prefer) Glute work early, especially if the knee can't tolerate much yet
Patellar taping with exercise B (may) Short-term pain relief to help you do the exercises
Running gait retraining B (may) Cadence, foot strike, or hip control cues, with multiple sessions
Foot orthoses, combined with exercise B (may) Cheap insoles can help, but only alongside strength work
EMG biofeedback on the inner quad Against Don't bother
Visual biofeedback on alignment Against Don't bother
Ultrasound, electrical stimulation, dry needling, manipulation Not helpful Should not be used for this

The single most important line in the whole document is this: exercise therapy combining hip- and knee-targeted work should be included, because it reduces pain and improves function in the short, medium and long term. Combining it with other things beats doing nothing, or doing inserts alone. Exercise is the critical ingredient, and everything else is a supporting act.

Note what's graded highest, too. Not stretching. Not foam rolling. Not the passive treatments that feel productive in a clinic. Strength.

The exercise program that works

The guidelines are specific about what "strength work" means for this condition, which is genuinely helpful because "strengthen your leg" is useless advice.

Hip-targeted work should target the posterolateral hip musculature. That's the glute complex: gluteus medius and minimus on the outside and back of the hip, plus the deep external rotators. The point of these muscles is femoral control. When they do their job, the thigh stays organised under load and the kneecap isn't asked to cope with a hip that's collapsing inward.

Knee-targeted work can be weight-bearing or not. Resisted squats and resisted knee extension both count. The guidelines say either works, which is a relief if your knee hates one of them. If deep squats are painful right now, start with seated knee extension and progress.

Combination beats knee-only work. This is graded A, and it's the difference between a program that works and one that plateaus. Hip and knee together, not one or the other.

A practical six-week structure looks like this, twice a week on non-consecutive days:

  • Weeks 1–2. Seated knee extension, 3 sets of 12–15 at a load where the last three reps are hard but pain stays under about 3 out of 10. Side-lying hip abduction, 3 sets of 12–15. Glute bridge, 3 sets of 12. Keep range shallow and comfortable.
  • Weeks 3–4. Add a banded squat or leg press to a depth that doesn't provoke pain, 3 sets of 8–10. Step-downs from a low box, 3 sets of 8 per leg. Add a lateral band walk for hip endurance, 2 sets of 15 steps each way.
  • Weeks 5–6 and beyond. Progress load, not just reps. Bulgarian split squats, heavier leg press, single-leg press, calf raises. This is the phase where tendon and bone actually adapt, and it's the phase most people skip.

Two rules that matter more than the exercise selection. First, progress load over time; the same three sets of twelve forever stops working. Second, keep the pain response tidy: discomfort during exercise up to a 3 or 4 out of 10 is acceptable if it settles within 24 hours and doesn't trend upward across the week.

This is a slow build. Expect meaningful change at six weeks and a real difference at twelve. The guidelines describe exercise as effective in the short, medium and long term, which is unusual in musculoskeletal medicine, and it's the reason the boring answer is the right one.

Should you keep running?

Usually, yes, and this is where a lot of people get bad advice in both directions. One camp says rest until it's completely gone, which deconditions you and removes the exact stimulus your knee needs to adapt. The other says push through anything, which is how a grumbly knee becomes a six-month problem.

The workable rule: run at a level of pain you can tolerate, where symptoms settle within a day, and where the weekly trend isn't getting worse. If your pain is a 3 during a run and a 1 the next morning, and that's been stable for two weeks while your strength work progresses, you're on track. If it's climbing and now waking you at night, you've overshot.

Practical levers, in order of how much they usually help:

  • Cut volume before you cut frequency. Three shorter runs beat two long ones, because frequency maintains the tissue's tolerance for load while volume comes down.
  • Flatten the route. Downhills load the patellofemoral joint hardest. Flat and rolling beats steep descents while you're symptomatic.
  • Soften the surface. A track, a treadmill, or a trail with some give is kinder than concrete.
  • Drop the speed work temporarily. Intervals and hill reps are the highest-load running you do. Park them for four to six weeks and keep easy running.
  • Watch the shoes. Very worn midsoles change how load is distributed. This isn't a cure, but a dead pair of shoes is a silly thing to be training in.

Gait retraining: modest, specific, evidence-backed

If pain persists despite good strength work, gait retraining is a reasonable next lever. The guidelines support cues to increase cadence, to shift a rearfoot striker toward a forefoot pattern, or to reduce peak hip adduction. A 2026 umbrella review of the gait retraining literature found the most consistent support for cadence increases of about 5 to 10%, plus softer-landing cues, both of which reduce the variables thought to drive patellofemoral load.

The 5–10% number is the useful part. If you run at 165 steps per minute, that's a target of roughly 173 to 182, achieved by taking shorter steps at the same pace, not by running faster. Most people find that a metronome app or a watch cadence field does the job for a few weeks until it becomes automatic.

What doesn't work is trying to consciously hold a perfect foot strike on every run forever. Gait retraining is a short block of deliberate practice, not a permanent vigilance project.

The stuff that doesn't work

Being direct about this saves people money and time.

Passive clinic treatments. Ultrasound, electrical stimulation, spinal manipulation and dry needling are not recommended for patellofemoral pain. They feel like treatment. They aren't treatment.

Biofeedback. Specifically, EMG biofeedback aimed at the inner quad, and visual biofeedback on limb alignment during exercise. Both were recommended against.

Chasing the inner quad. The idea that the vastus medialis obliquus is asleep and needs to be woken up to "pull the kneecap back into place" is one of the most durable myths in running injury. The guidelines explicitly advise against biofeedback for it, and kneecap tracking as a mechanical story doesn't hold up the way it was taught.

Stretching as the main intervention. It's not harmful, and the guidelines allow lower-limb stretching as part of a combined program. But it is not what drives recovery, and if it's all you do, nothing changes.

Waiting for it to disappear on its own. The consensus literature describes this as a condition that frequently persists. Time alone isn't a treatment plan.

When it's probably not runner's knee

Patellofemoral pain is diffuse and around or behind the kneecap, worse with loaded bending, and usually without swelling. Some things that look similar need a different approach:

  • Pain on the outside of the knee, especially at a consistent mileage point. That's more typical of iliotibial band-related pain than patellofemoral pain.
  • Pain right at the bony point below the kneecap, worse at the start of a run and easing with warm-up. That pattern suggests patellar tendinopathy, which has a different loading protocol.
  • Locking, catching, giving way, or a joint that swells within hours. Get assessed. That's not this.
  • Night pain that wakes you, or pain with fever. Also get assessed, promptly.

The guidelines note that clinicians diagnose this by reproducing pain with movements like squatting, after ruling out IT band pain and patellar tendinopathy. That's the differential, and it's worth knowing because the treatments differ.

What your watch can actually help with

Your wearable won't diagnose a knee. It will, however, tell you whether the load story makes sense, and that's the part you can control.

Training load trends. A jump in load in the weeks before symptoms started is the single most common pattern. If you're on watchOS 11 or later, the Training Load view gives you a rolling comparison against your 28-day baseline. A jump from well below to well above your usual range in two or three weeks is exactly the kind of ramp that produces overuse complaints.

Volume and frequency, not just intensity. Load isn't only about hard sessions. Adding a fourth run day and 15% more distance in the same fortnight is a meaningful change even if every run was easy.

Cadence. Many watches display steps per minute in the workout summary, and it's the cheapest feedback loop for the gait-retraining piece above.

Recovery markers as a sanity check. Not because a low HRV means your knee hurts, but because poor sleep and high overall strain change how much load you can absorb. If your resting heart rate is up 4 bpm and your HRV has been low for a week, that's a bad week to also add mileage.

The trend, not the day. A knee that's mildly sore after a hard week and fine after an easy one is behaving normally. A knee that's sore every week regardless of load is telling you the capacity problem hasn't been solved yet.

This is the kind of thing Century AI is built for: a daily health and recovery score from your HRV, resting heart rate and sleep, so you can see a load spike or a recovery dip before it turns into a missed month of running.

A six-week plan

If you want something to actually follow:

  1. Week 1. Two strength sessions, hip and knee. Keep running, cut volume by 20–30%, remove hills and speed work. Record pain during and 24 hours after each run.
  2. Week 2. Same strength sessions, slightly more load. Hold running steady. If pain is stable or improving, continue.
  3. Week 3. Add a third strength session if recovery allows. Start a cadence block: three short runs a week with a 5–7% cadence increase.
  4. Week 4. Progress strength to heavier, lower-rep work. Begin adding volume back at 10% per week, flat routes only.
  5. Week 5. Reintroduce one light quality session, such as strides or short flat intervals, if pain hasn't increased.
  6. Week 6. Reassess honestly. Pain with stairs, pain with sitting, pain during running, and how much running you're doing. If two of those improved, keep going. If none did, get a physio assessment.

Two things to expect. It will feel too easy in weeks one to three, and it will feel unglamorous throughout. That's the shape of rehab that works.

The short version

  • Runner's knee is patellofemoral pain, defined by pain around or behind the kneecap that's worse with loaded knee bending. Imaging doesn't diagnose it.
  • It's a load-tolerance problem, not a structural defect. Knee strength predicts it; knee alignment mostly doesn't.
  • Combined hip and knee strengthening is the highest-graded treatment, and the only one that matters much. Six to twelve weeks, progressive load.
  • Keeping running at tolerable pain levels is usually fine, and better than total rest, as long as the weekly trend isn't worsening.
  • A 5–10% cadence increase and softer landings have the best gait-retraining evidence.
  • Passive treatments, inner-quad biofeedback and IT band folklore don't work. Skip them.
  • If you have locking, giving way, or rapid swelling, this isn't the right article. Get assessed.

Most runners with patellofemoral pain get better. The ones who get better fastest are usually the ones who started lifting twice a week and kept running easy, rather than the ones who stopped everything and waited.

If you're dealing with a different running niggle, the guide to shin splints and how to keep running covers the same load-management logic, and how to increase running mileage without getting hurt is the ramp-rate companion piece.

This isn't medical advice. Knee pain has several causes that need a proper assessment, so if it's not settling, see a physio or a doctor.


Century AI helps you understand your body with a daily health score, recovery score, and sleep insights — using the watch you already wear.

Tagsrunners-kneepatellofemoral-painrunninginjuryknee-painstrength-trainingrehabtraining-load

Wearable measurements are estimates, not diagnoses. Read Century’s methodology and medical disclaimer.