Shoulder MRI: what it finds that an X-ray misses (2026 UK guide)

The one-line answer

If your shoulder pain has lasted more than six weeks, is worse at night, or started after a fall or dislocation, an X-ray will almost certainly be normal and unhelpful. A shoulder MRI is the scan that actually sees the rotator cuff, the labrum, the tendons and the bursa - the four structures where nearly every real shoulder diagnosis lives.

The rest of this guide is what the MRI actually finds, when a plain MRI is enough, when you need an MRI arthrogram instead, and how the diagnosis tends to differ if you are under 40 versus over 40. It's written for the patient who has already had the X-ray, been told it looks fine, and is still in pain.

The four things an X-ray of the shoulder cannot see

An X-ray of the shoulder is a two-second, low-cost look at bone. It's genuinely useful for one thing: ruling out a fracture, dislocation or advanced arthritis. Beyond that, it is silent on almost every soft-tissue structure in the joint. The four blind spots matter:

• The rotator cuff. The four tendons (supraspinatus, infraspinatus, teres minor, subscapularis) that lift and rotate the arm. An X-ray shows none of them. A cuff tear looks identical to a normal shoulder on an X-ray.
• The labrum. The fibrocartilage rim around the socket that keeps the ball of the shoulder in place. SLAP and Bankart tears are entirely invisible to X-ray.
• The tendons and bursa. Biceps tendinopathy, subacromial bursitis and calcific tendonitis rarely show on X-ray. Occasionally you can see a large calcific deposit, but even then you cannot tell if it is the pain source.
• Joint fluid and inflammation. An effusion, capsulitis, or an inflamed bursa are all soft-tissue changes. MRI reads them clearly. X-ray does not.

That's why the standard sequence in a shoulder pain workup is X-ray first (fast, cheap, rules out bone problems) and then MRI if the pain persists or the mechanism suggests soft-tissue injury. Skipping the MRI is why so many shoulder problems get treated as "just tendinopathy" for a year before someone finally scans it properly.

The seven diagnoses shoulder MRI is built for

These are the seven conditions that a good shoulder MRI report will confidently confirm or exclude. Between them, they account for the overwhelming majority of persistent shoulder pain in UK adults.

• Partial-thickness rotator cuff tear. The tendon is frayed or partially torn but not through and through. MRI shows the depth, location and which of the four cuff tendons is involved. This information changes whether the treatment is physio, injection or surgery.
• Full-thickness rotator cuff tear. The tendon is torn all the way through. MRI shows tear size, degree of retraction (how far the torn end has pulled back) and whether the muscle has started to atrophy or fill with fat. All three affect whether the tear is repairable.
• Subacromial bursitis. Inflammation of the fluid-filled sac that sits above the cuff. Very common, often the source of the classic "painful arc" between 60 and 120 degrees of arm elevation.
• SLAP lesion. A tear of the superior labrum where the biceps tendon attaches. Typical in throwing athletes and after falls onto an outstretched arm. Best seen on an MRI arthrogram.
• Bankart lesion. A tear of the anterior-inferior labrum, almost always after a shoulder dislocation. Also best seen with contrast.
• Frozen shoulder (adhesive capsulitis). The joint capsule thickens and contracts, dramatically limiting movement. Often clinically obvious, but MRI helps rule out cuff tears and shows the classic capsular thickening.
• Calcific tendonitis and biceps tendon rupture. Calcium deposits within the cuff or a torn long head of biceps ("Popeye deformity") - MRI localises the deposit or confirms the rupture and its retraction.

MRI vs MRI arthrogram - which do you need?

This is the single most under-explained bit of shoulder imaging in the UK, and the reason a lot of scans get repeated. There are two studies, and they answer different questions.

A standard shoulder MRI is a non-contrast scan. You lie in the scanner for about 25 minutes and the machine takes pictures of the soft tissues around the joint. It is very good for the rotator cuff, the tendons, the bursa, the bone marrow and for frozen shoulder. If the clinical question is "is there a rotator cuff tear", a standard MRI is almost always the right test.

An MRI arthrogram adds a small amount of contrast injected into the joint by a radiologist under X-ray or ultrasound guidance, immediately before the scan. The contrast distends the capsule and outlines the labrum, which is otherwise a thin structure pressed flat against the socket. If the clinical question is "is there a SLAP tear" or "is there a Bankart lesion after that dislocation", an arthrogram is significantly more accurate than a plain MRI.

If you're not sure which you need, don't guess. A shoulder consultant or MSK physiotherapist will make the call in a five-minute assessment, and picking the right scan first time saves you the cost of two scans.