How to Diagnose Shoulder Pain: Exams, Tests & Imaging

Diagnosing shoulder pain relies on a layered process that starts with your history and a hands-on physical exam, then moves to imaging only when the clinical picture warrants it. No single test, whether performed in the exam room or the radiology suite, reliably identifies every shoulder condition on its own. Instead, clinicians combine clusters of physical maneuvers with targeted imaging to narrow down the source, which is usually the rotator cuff or the subacromial bursa but can also be the labrum, the joint capsule, a nerve, or even the cervical spine.

Why Your History Matters More Than You Think

Before anyone touches your shoulder, the conversation you have with your clinician does real diagnostic work. A systematic review found that specific history elements shift the probability of particular diagnoses in meaningful ways. For instance, a rotator cuff tear becomes more likely if you are over 60, have done a lot of overhead work or heavy lifting, or have a family member with a rotator cuff problem. Adhesive capsulitis (frozen shoulder) is more likely in people with diabetes or thyroid disorders. Glenohumeral arthritis is more likely if you have a history of prior dislocation, are over 75, or already have knee osteoarthritis. And posterior labral tears are more common among football players.¹ These are not iron-clad rules, but they tell a clinician where to focus the physical exam and what imaging to order if needed.

The character of the pain matters, too. Pain that radiates below the elbow, worsens with neck movement, or comes with tingling or numbness raises a flag that the problem may originate in the cervical spine rather than the shoulder itself. Pain that is worst at night and disturbs sleep is a classic rotator cuff complaint. Sudden onset after trauma obviously points toward an acute injury, while insidious stiffness developing over weeks suggests frozen shoulder or early arthritis.

Rotator Cuff Examination

The rotator cuff is the most common source of shoulder pain. In one prospective study of over 200 patients with shoulder pain in primary care, the rotator cuff showed the highest rate of pathology on imaging, appearing in about half of patients on ultrasound and roughly two-thirds on MR arthrography.² Clinicians use a battery of physical tests to evaluate it, and the evidence on which ones perform best is surprisingly nuanced.

Among pain-provocation tests, the painful arc (lifting your arm out to the side and noting pain in a specific mid-range zone) stands out. A systematic review in JAMA found it was the only pain-provocation maneuver with a meaningfully elevated positive likelihood ratio for rotator cuff disease, meaning a positive result roughly triples the odds of a rotator cuff problem.³ Among strength tests, the lag signs are the heavy hitters. The external rotation lag sign, where the examiner holds your arm in a rotated position and lets go to see if it drops, had the highest diagnostic odds ratio for rotator cuff tears in a recent meta-analysis.⁴ The internal rotation lag sign performed similarly well, and a normal result on that test was the most accurate finding for ruling out a full-thickness tear.³

Jobe’s test (also called the empty can test), where you resist downward pressure with your arm at shoulder height and your thumb pointing down, is widely taught and frequently used. It is quite sensitive for supraspinatus tears, catching about 88% of them, but its specificity is only around 62%, meaning a fair number of people without a tear will still test positive.⁵ In practice, that means a positive Jobe’s test should prompt further evaluation rather than a definitive diagnosis.

Impingement Tests and Their Limits

The Neer sign and the Hawkins-Kennedy test are the two classic maneuvers for subacromial impingement, where the rotator cuff or bursa gets pinched beneath the bony arch of the shoulder. Both involve the examiner moving your arm into positions that reproduce that pinching effect. They are sensitive tests, meaning they are good at catching a problem when it exists. The Hawkins sign has a sensitivity of about 88–92% for cuff tearing and bursitis, and the Neer sign runs around 75–85%.⁶ When both tests are negative, the combined negative predictive value is high, around 96% for bursitis and 90% for rotator cuff tearing, so two normal impingement tests go a long way toward ruling out those conditions.

The catch is specificity. A meta-analysis of impingement tests found that the Hawkins-Kennedy, Neer, and empty can tests all have specificities in the range of 57–62%.⁷ That means roughly four out of ten people who test positive do not actually have impingement. The drop arm test and lift-off test flip this picture: their specificities are very high (92–97%), but their sensitivities are low (21–42%), making them useful for confirming impingement when positive but poor at detecting it when negative. Clinicians typically use the sensitive tests first to screen and then apply the more specific tests to firm up the diagnosis.

There is an additional wrinkle. A study using arthrographic confirmation found that diagnostic injections into the subacromial space did not reduce pain significantly more than injections that ended up in the deltoid muscle instead. Pain relief occurred regardless of where the anesthetic landed, which calls into question how much a positive “impingement test” truly localizes the problem to the subacromial space.⁸ This is part of why the term “subacromial impingement syndrome” has fallen somewhat out of favor and why many clinicians now prefer broader descriptions like “subacromial pain.”

Testing for Labral Tears

Labral tears, especially the type called SLAP lesions (involving the top part of the socket where the biceps tendon attaches), are notoriously difficult to diagnose with physical examination alone. The Speed test and Yergason test, two maneuvers traditionally used for biceps and labral problems, perform poorly. When compared against arthroscopy, the confidence intervals for their likelihood ratios spanned 1, meaning a positive or negative result did not meaningfully change the odds of having the condition.⁹

No single physical test is both highly sensitive and highly specific for a SLAP lesion.¹⁰ The workaround is test combinations. A meta-analysis found that the O’Brien test and crank test combination was the most sensitive two-test pair, while the Yergason and anterior slide test combination was the most specific.¹¹ Another approach that showed promise was combining two or three sensitive tests (like the O’Brien, apprehension, or compression-rotation test) with one specific test (like the Yergason or biceps load II test), which pushed sensitivity and specificity to approximately 70% and 95%.¹⁰ Labral tears are ultimately one of the diagnoses where imaging, particularly MR arthrography, often plays a decisive role.

Instability Assessment

If your shoulder has ever slipped out of place or you feel a sense of looseness or “dead arm” episodes, your clinician will run a different set of tests. The apprehension test (moving your arm into a cocked throwing position to see if it provokes fear of dislocation), the relocation test (applying backward pressure in that same position to see if the apprehension resolves), and the release or surprise test (suddenly letting go of that backward pressure) form the standard trio. The anterior release test demonstrates the best overall combination of sensitivity and specificity for diagnosing anterior instability.¹²

One detail that makes a real difference in the exam room: the criterion for calling a test “positive” matters enormously. When the examiner scores these tests based on whether the patient feels apprehension (that sense of impending dislocation), the apprehension test achieved a specificity of 96% and the relocation test about 92%. When the criterion was simply pain, accuracy dropped across the board.¹³ Inter-examiner reliability data confirms this same pattern: agreement between different examiners was best when apprehension was used as the diagnostic criterion and was poor when pain was used instead.¹⁴ If a clinician asks whether the maneuver feels like your shoulder is about to come out rather than just asking whether it hurts, that is a sign of a thorough evaluation.

Frozen Shoulder and Range of Motion

Adhesive capsulitis, commonly called frozen shoulder, is one of the few shoulder diagnoses where the physical exam is often sufficient on its own. The hallmark is a global loss of both active and passive motion, meaning you cannot move the shoulder fully, and neither can anyone else move it for you. Diagnostic criteria typically require limitation in at least two directions of passive motion lasting more than a month, with no abnormalities on plain X-rays.¹⁵ External rotation is usually the most restricted. If your shoulder hurts and feels stiff but someone can passively move it through its full range, frozen shoulder is unlikely.

When the Pain Is Not Actually Coming From the Shoulder

One of the trickiest diagnostic challenges is the shoulder that hurts because of a cervical spine problem. Nerve root compression in the neck can produce pain that feels indistinguishable from a rotator cuff tear. Two newer clinical tests aim to sort this out. The Arm Squeeze Test, which involves squeezing the middle third of your upper arm and comparing the pain to squeezing around the shoulder itself, was positive in about 97% of patients with cervical nerve root compression but in fewer than 4% of patients with rotator cuff tears.¹⁶ The Swimmer Arm-to-Shoulder test, a more recent addition, showed similar promise with a sensitivity of about 89% and specificity of about 96% for distinguishing shoulder impingement from cervical radiculopathy.¹⁷

Beyond the cervical spine, shoulder pain can occasionally be referred from entirely non-musculoskeletal sources. Cardiac problems, liver disease, lung conditions, and even kidney stones can produce pain in the shoulder region. A clinician screening for red flags will typically ask about chest symptoms, shortness of breath, unexplained weight loss, and whether the pain relates to exertion or meals in ways that do not match a mechanical shoulder pattern.

When Nerve Injury Is the Culprit

In a small but meaningful subset of patients, nerve damage contributes to shoulder pain and weakness. A prospective electromyography (EMG) study of patients with massive rotator cuff tears found that about 12% had neurologic lesions on testing, including suprascapular neuropathy, cervical root problems, and axillary nerve palsy.¹⁸ EMG and nerve conduction studies are typically reserved for cases where weakness seems out of proportion to the structural damage seen on imaging, or when muscle wasting is visible. One caveat: EMG has limited diagnostic value in the first few weeks after acute injury because nerve-related changes take time to develop on testing.¹⁹

Imaging: Starting With X-Rays

Plain radiographs are almost always the first imaging study. The American College of Radiology guidelines recommend X-rays as the initial imaging modality for both acute and chronic shoulder pain.²⁰²¹ X-rays are inexpensive, widely available, and fast. They show fractures, arthritis, calcific tendinitis, and bony abnormalities of the acromion. A standard shoulder X-ray series includes several views taken in different arm positions. What X-rays cannot show well is soft tissue: tendons, the labrum, and cartilage are essentially invisible on plain films. For that reason, X-rays are a starting point, not an endpoint, when soft-tissue injury is suspected.

Ultrasound Versus MRI for the Rotator Cuff

For rotator cuff tears specifically, ultrasound and MRI perform remarkably similarly. A meta-analysis comparing the two found no significant difference in either sensitivity or specificity for diagnosing partial- or full-thickness tears.²² Another study reported near-identical accuracy for full-thickness tears (98% for ultrasound, 100% for MRI) and similar accuracy for partial-thickness tears (87% versus 90%).²³

So why does MRI seem to get ordered more often? Ultrasound is operator-dependent, meaning the quality of the exam hinges on the skill and experience of the person holding the probe. It also has a limited view of deeper structures like the labrum and the undersurface of certain tendons. MRI, by contrast, provides a comprehensive look at the entire shoulder in a single session, including bone marrow, cartilage, and structures that ultrasound cannot easily reach. Ultrasound has advantages of its own: it is cheaper, does not require lying still in a tube, allows the examiner to test the shoulder dynamically (watching tendons move in real time), and can be done in the office during the same visit. For a straightforward rotator cuff question in a patient who does not need labral evaluation, ultrasound in experienced hands is a perfectly valid choice.

MR Arthrography for Labral and Subtle Injuries

Standard MRI is good at many things, but it has a genuine weakness when it comes to labral tears and partial-thickness tears on the joint side of the rotator cuff. MR arthrography, where contrast dye is injected into the joint before scanning, fills that gap. At 3 Tesla field strength, MR arthrography detected SLAP tears with 98% sensitivity compared to 83% for conventional MRI. For anterior labral tears, the numbers were 98% versus 83%. For partial-thickness articular surface tears of the supraspinatus, the gap was dramatic: 97% sensitivity with arthrography versus 68% without it.²⁴ The injected contrast distends the joint and outlines the labrum and undersurface of the tendons, making subtle tears visible that would otherwise hide on a standard scan.

MR arthrography is not always necessary. It adds an invasive step (the injection itself, which can cause brief discomfort), takes more time, and costs more. It is most valuable when labral pathology or subtle partial-thickness rotator cuff tears are the clinical question, particularly in younger or athletic patients where surgical planning hinges on the precise location and extent of a labral tear.

CT Scans and Bone Loss in Shoulder Instability

CT scanning has a specific niche in shoulder diagnosis: measuring bone loss in patients with recurrent instability. When the shoulder dislocates repeatedly, it can erode bone from the front of the socket (the glenoid). The amount of bone loss is one of the major factors that determines whether a soft-tissue repair like a Bankart procedure will be enough or whether a bone-grafting procedure is needed. Three-dimensional CT is currently the most widely used method for this measurement.²⁵ A systematic review found that CT-based measurements were accurate in the vast majority of studies (about 80–86% depending on the reconstruction method), whereas plain X-rays were accurate in none of the studies examined.²⁶ One study showed that a glenoid index calculated from 3D CT accurately predicted the need for bone grafting in 96% of patients.²⁷

Why Clinicians Combine Tests Rather Than Rely on One

A recurring theme across the research is that no single exam maneuver nails the diagnosis on its own. The evidence consistently shows that clustering tests improves accuracy. For subacromial pain, one study found that using a battery of five tests and defining a positive result as three or more being positive yielded a significant area under the curve of 0.79.²⁸ The same principle applies to labral testing, instability testing, and rotator cuff testing: combining a sensitive test (one that catches most true cases) with a specific test (one that reliably rules out non-cases) produces far better diagnostic accuracy than using either one in isolation.

This is worth knowing because patients sometimes worry when a clinician runs them through a long series of maneuvers. It does not mean the examiner is confused or fishing. It means they are building a probability picture, layering clues the way a detective layers evidence. Each test shifts the probability up or down, and the pattern of results across all tests is far more informative than any individual positive or negative.

Patient-Reported Outcome Questionnaires

Alongside physical examination and imaging, clinicians increasingly use standardized questionnaires to gauge how your shoulder problem affects daily life. The most commonly used tools include the DASH (Disabilities of the Arm, Shoulder and Hand), the SPADI (Shoulder Pain and Disability Index), and the ASES (American Shoulder and Elbow Surgeons) score. These questionnaires have been shown to have excellent test-retest reliability, meaning your score stays consistent if your condition has not changed.²⁹ A systematic evaluation rated the ASES, the Simple Shoulder Test, and the Oxford Shoulder Score as the best-validated instruments available.³⁰ These scores do not diagnose the structural problem, but they quantify your functional impairment in a way that helps guide treatment decisions and track progress over time.

Shoulder Pain in Young Athletes

The diagnostic approach shifts when the patient is a child or teenager, because the growing skeleton has structures that adults do not: growth plates. In skeletally immature athletes, especially overhead throwers, repetitive stress on the proximal humeral growth plate (sometimes called “Little Leaguer’s shoulder”) is an important diagnosis to catch early, because delayed recognition can lead to lasting complications.³¹ The spectrum of injuries in adolescents can be distilled into recognizable patterns, but the evaluation requires attention to growth-plate status on imaging and a high index of suspicion for overuse injuries that adults simply cannot get.³² X-rays are particularly important in this age group because they show growth-plate widening or irregularity, and MRI may be added if a labral tear or stress fracture is suspected.