The plain answer: X-ray first, MRI when it matters
If you have knee pain and you are trying to work out which scan you actually need, the plain answer is this. In almost every case the sensible first step in the UK in 2026 is a weight-bearing X-ray of the knee. It is quick, cheap, low-dose and tells you whether there is a fracture, whether the joint is out of alignment, and whether the joint space is narrowing from osteoarthritis. That single image answers the question for a very large share of patients over 45. An MRI is added when the X-ray is normal but the symptoms are not, when the mechanism of injury was twisting rather than impact, or when the pain is behaving mechanically - locking, catching, giving way.
The reason people get this the wrong way round is that MRI feels like the "better" scan, and in image terms it is. It sees soft tissue an X-ray cannot see. But an MRI ordered before an X-ray, in a patient who turns out to have straightforward osteoarthritis, is a slower and more expensive route to the same answer, and it often generates incidental findings that confuse the picture more than they clarify it. The right sequence for the knee is almost always X-ray first, then MRI if the X-ray does not fit the story.
What an X-ray of the knee actually shows
A knee X-ray is a two or three-view study, typically weight-bearing, that takes ten minutes from walking in to walking out. It answers a small number of very important questions extremely well.
- Fractures. Any suspected break of the femur, tibia, fibula or patella after a fall, direct impact or road-traffic mechanism is an X-ray question first.
- Alignment. Genu varum (bow legs) or valgum (knock knees), patellar tracking, tibial plateau tilt. These matter for surgical planning.
- Joint-space narrowing. The hallmark of osteoarthritis. Weight-bearing views are essential - non-weight-bearing X-rays systematically under-read the degree of OA.
- Osteophytes and subchondral sclerosis. Bony spurs and bone hardening that accompany established OA.
- Bone lesions and tumours. Rare, but X-ray remains the appropriate first-line screen for suspicious bony changes.
- Loose bodies. Small fragments of bone or ossified cartilage floating in the joint that can cause locking.
What an X-ray cannot show you is almost everything soft - meniscus, ligaments, tendons, cartilage surface, bone-marrow oedema, joint effusion. If the pain feels like it is from one of those, or if the X-ray is normal but the knee is not, that is when MRI enters. See our MRI overview for how the scan itself works.
What an MRI of the knee shows that X-ray misses
A knee MRI is a 20 to 30 minute scan on a 1.5T or 3T scanner that produces multi-planar images sensitive to water, fat and structure. What that means in practice is that everything soft becomes visible.
- Meniscal tears. The medial and lateral menisci - the two crescent-shaped shock absorbers - are the single most common finding on knee MRI. Type, location, displacement and root involvement are all visible.
- Ligaments. ACL, PCL, MCL and LCL are all clearly seen, with tears graded from sprain through partial to complete rupture.
- Articular cartilage. Chondral defects, thinning and full-thickness loss. This matters for cartilage-preserving surgery decisions.
- Bone marrow oedema. Bruising inside the bone that is invisible on X-ray. Often the true source of pain in occult stress reactions or after impact.
- Joint effusion. Excess fluid, which points at inflammation, infection or internal derangement.
- Baker's cyst. A fluid-filled swelling at the back of the knee, often secondary to intra-articular pathology.
- Plica, patellar tendinopathy, quadriceps tendon injury. Extensor mechanism pathology that no X-ray will ever reveal.
Our musculoskeletal MRI page explains how a subspecialist MSK radiologist reads these findings differently to a general radiologist. For the knee the difference genuinely matters.
Symptom-by-symptom: which do you need?
The most useful way to decide is by what the knee is actually doing, not by the diagnosis you suspect. Use this as a starting point, and confirm the plan with a clinician.
| What your knee is doing | First scan | Why |
|---|---|---|
| Trauma with inability to bear weight or obvious deformity | X-ray urgently | Rule out fracture. MRI later if X-ray normal. |
| Twisting injury with immediate swelling within hours | MRI (X-ray if bony tenderness) | Classic ACL or meniscus pattern. Soft-tissue answer needed. |
| True mechanical locking (knee stuck) | MRI | Displaced meniscal tear or loose body. Both soft or small. |
| Giving way on stairs or turning | MRI | Ligament or meniscus instability. X-ray will look normal. |
| Gradual ache over months, worse on stairs, age 45+ | Weight-bearing X-ray | Osteoarthritis is the leading candidate. X-ray answers it. |
| Anterior knee pain in a runner, age under 40 | Neither initially | Patellofemoral pain syndrome. Physiotherapy first. |
| Painless clicking only | Neither | Extremely common. Not a scan question. |
| Persistent pain 4-6 weeks after normal X-ray | MRI | The X-ray has done its job and ruled out bone. |
Under-45 vs over-45 knee pain
Age changes the pre-test probability of what is going on inside a knee, and that changes the right first scan. It is a general pattern, not a rule, but it helps direct the decision.
Under 45. The tissues that fail in this age group are usually soft. Acute knee pain in a runner, footballer, skier or gym-goer is far more often a meniscal or ligament story than an OA story. Weight-bearing X-rays are usually normal. An MRI is the higher-yield scan for a mechanical or twisting injury in this group, and if the mechanism is a clear pivot with immediate swelling, going straight to MRI is often reasonable.
Over 45. The tissues that fail begin to include the joint surface itself. Chronic mechanical pain, morning stiffness lasting less than 30 minutes, and pain worse on stairs strongly points at osteoarthritis, which a weight-bearing X-ray diagnoses cheaply and reliably. Going straight to MRI in this group is where over-imaging tends to happen, and where the next section becomes important.
The "meniscus tear on MRI" overdiagnosis problem
This is the part almost no other UK knee blog will tell you, and it is the single most useful thing to understand before booking an MRI.
Meniscal tears are extremely common on MRI in people who have no knee pain at all. Published radiology series show that in adults over 50 without knee symptoms, roughly one in three MRIs shows a meniscal tear. Over 65, that number climbs higher still. The tear is real. It is just often not the source of the symptoms - it is a bystander finding in a degenerating joint.
Practically, what that means is: if an over-50 patient with gradual medial-side knee pain has an MRI before a weight-bearing X-ray, and the report says "medial meniscal tear", it is very easy for that tear to be labelled the cause, and for the patient to end up on an arthroscopy waiting list. The randomised evidence for arthroscopic partial meniscectomy in degenerative tears without true mechanical locking is genuinely poor. In many patients the true driver is early osteoarthritis, and the treatment is physiotherapy, weight optimisation and time - not surgery on a tear that would have been silent forever if nobody had looked for it.
Imaging is only useful if it changes what you do next. If the answer is going to be conservative management either way, the scan may not be earning its place.
None of this argues against MRI. MRI is the right tool for the right patient. It argues for X-ray first in the age group where OA is likely, and for interpreting an MRI report in the context of the patient's actual symptoms rather than in isolation.
