What Tests Are Used to Diagnose a Torn Rotator Cuff?

Diagnosing a torn rotator cuff involves a layered approach that starts with hands-on clinical tests in the exam room and escalates to imaging when needed. No single test confirms a tear on its own, so clinicians combine findings from physical maneuvers, plain X-rays, ultrasound, or MRI to piece together the diagnosis. The process is more nuanced than most people expect, because a torn rotator cuff can look very different depending on which tendon is involved, how deep the tear goes, and whether it is actually causing your symptoms.

Physical Examination Tests

The first round of testing happens with your doctor’s hands and your shoulder movements. Physical exam maneuvers fall into two broad categories: pain provocation tests, which check whether specific positions reproduce your pain, and strength tests, which assess whether a tendon is too damaged to generate normal force. Both categories matter, but strength deficits tend to be more telling. A systematic review in JAMA found that among pain provocation tests, the painful arc test was the single most useful finding for identifying rotator cuff disease, while among strength tests, the external rotation lag sign and internal rotation lag sign were the most accurate for detecting full-thickness tears.¹

Which specific maneuvers your doctor chooses depends on which tendon they suspect. The supraspinatus, the most commonly torn rotator cuff tendon, is tested with the empty can test (also called the Jobe test) and the full can test. In both, you hold your arms out in front of you at an angle while the examiner pushes down, but the hand position differs slightly. Research shows the two are roughly equivalent in accuracy when the examiner interprets weakness rather than pain as the sign of a tear, with the full can test reaching about 75% accuracy and the empty can test about 70%.² A later study comparing both tests against surgical findings confirmed a significant correlation for each, with the full can test performing slightly better overall.³

The drop arm test is another common maneuver for the supraspinatus. The examiner raises your arm overhead and asks you to lower it slowly. If the arm drops suddenly or you cannot control the descent, that suggests a significant tear. This test has very high specificity, meaning a positive result is fairly reliable, but its sensitivity is low, so a negative result does not rule a tear out.⁴

Testing the Subscapularis

The subscapularis sits on the front of the shoulder blade and is the most powerful internal rotator of the shoulder. Because its tears are harder to detect clinically and on imaging, several dedicated maneuvers exist. The lift-off test, the oldest of these, asks you to place the back of your hand against your lower back and push away from it. If you cannot generate force, the subscapularis is likely torn. A systematic review identified the lift-off test, the internal rotation lag sign, the belly-press test, and the Napoleon test as the main clinical tools for this tendon.⁵

Each of these tests has trade-offs. In one study comparing all four, the belly-press test was the most sensitive, picking up the highest proportion of actual tears, while the lift-off test was the most specific, reaching 100% specificity for any subscapularis tear.⁶ In practical terms, that means a positive lift-off test is a strong indicator that a tear exists, but a negative belly-press test is less reassuring that the tendon is intact.

Plain X-Rays and What They Show

X-rays cannot show a rotator cuff tear directly because tendons do not appear on standard radiographs. So why do doctors order them? X-rays reveal indirect clues and help rule out other causes of shoulder pain such as fractures, arthritis, or bone spurs. One key measurement is the acromiohumeral distance, the gap between the top of the arm bone and the bony arch above it. In a healthy shoulder, this gap averages around 10 mm. When it narrows to 7 mm or less, that suggests a large rotator cuff tear has allowed the humeral head to migrate upward.⁷ Reduced acromiohumeral distance has also been correlated with the size of the tear and the degree of fatty degeneration in the rotator cuff muscles on MR arthrography.⁸

The Society of Radiologists in Ultrasound has published consensus imaging algorithms recommending that plain radiographs be included as a baseline step in evaluating suspected rotator cuff disease, before any advanced imaging.⁹ Think of X-rays as the screening layer: they set the stage for a more targeted workup rather than providing the diagnosis themselves.

Ultrasound

Musculoskeletal ultrasound has become a first-line imaging tool for rotator cuff tears in many practices. It is fast, does not involve radiation, allows the examiner to move the shoulder in real time, and costs substantially less than MRI. A large systematic review and meta-analysis found that ultrasound had a median accuracy of about 93% for full-thickness supraspinatus tears, dropping to around 81% for partial-thickness tears.¹⁰ That gap between full-thickness and partial-thickness detection is worth remembering: ultrasound excels at catching complete tears but is less reliable when only part of the tendon is damaged.

A major caveat is operator dependence. The quality of an ultrasound exam depends heavily on the skill and experience of the person holding the probe. In the hands of a fellowship-trained musculoskeletal radiologist or a surgeon who performs shoulder ultrasound routinely, accuracy is high. In less experienced hands, tears can be missed or over-called. If your doctor orders an ultrasound, it is reasonable to ask whether the person performing it specializes in musculoskeletal imaging.

MRI

Magnetic resonance imaging is the most widely used advanced imaging test for suspected rotator cuff tears and gives the most detailed view of soft tissues around the shoulder. Beyond simply detecting a tear, MRI shows how large it is, which tendons are involved, whether the torn edges have retracted, and whether the rotator cuff muscles have undergone fatty infiltration, a change that can affect surgical outcomes. Radiologists assess muscle quality using grading systems such as the Goutallier classification, the tangent sign, and the fish backbone sign, all of which help predict whether a tendon repair is likely to heal.¹¹

For detecting subscapularis tears specifically, two MRI readers in one study achieved sensitivities of 85% to 89% and specificities of 70% to 75%, with substantial agreement between them.¹² These numbers are good but not perfect, which is why a surgeon sometimes discovers a tear intraoperatively that the MRI missed, or vice versa.

Where standard MRI struggles most is partial-thickness tears. The edges of a partial tear can be subtle, and fluid signals that help light up full-thickness tears on MRI sequences are less pronounced. One review noted that imaging studies have been shown to have accuracy as low as 63% to 75% for diagnosing partial-thickness tears.¹³

MR Arthrography and CT Arthrography

When standard MRI leaves questions unanswered, particularly about partial-thickness tears or labral damage, the next step is often MR arthrography. This involves injecting contrast dye into the joint before scanning, which outlines the internal structures more crisply. Somewhat counterintuitively, a meta-analysis comparing standard MRI and MR arthrography for partial-thickness tears found that conventional MRI actually had higher pooled sensitivity (about 70%) than MR arthrography (about 45%), though MRI also had higher specificity (roughly 95% versus 76%).¹⁴ These results raise the interesting question of whether the added discomfort of a joint injection always pays off diagnostically. The answer depends on the clinical scenario: MR arthrography remains valuable for certain labral and capsular pathology that standard MRI misses.

In adolescents, MR arthrography has been found to be only 44% sensitive for partial rotator cuff tears, with 87% specificity.¹⁵ The developing shoulder can be trickier to read on imaging, which is worth knowing if you are the parent of a young athlete with shoulder pain.

CT arthrography is a separate option that combines a contrast injection with a CT scan instead of an MRI. In patients who cannot undergo MRI (because of certain metal implants or claustrophobia), CT arthrography provides an alternative with strong accuracy. One study of non-operated shoulders reported sensitivity and specificity ranging from 87% to 100% when compared against arthroscopy, and in previously operated shoulders, CT arthrography reached 94% accuracy compared with just 25% for MRI.¹⁶ That last number highlights how difficult post-surgical shoulders are to read on standard MRI, where metal artifact and scar tissue can obscure the picture.

The Role of Diagnostic Injections

Sometimes shoulder pain is so severe that a patient cannot cooperate with physical exam maneuvers, making strength testing unreliable. In these situations, a subacromial lidocaine injection can temporarily eliminate pain and allow the examiner to retest. Numbing the area does change the test results: sensitivity tends to drop (because a patient who was “weak from pain” may test strong once the pain is gone), while specificity improves (because remaining weakness after pain relief is more likely to reflect a true structural tear). One study found that after lidocaine injection, the drop arm test reached a specificity of 96%.¹⁷ Another prospective study confirmed the same pattern across multiple lag sign tests.¹⁸

The injection itself is not the diagnostic test; it is a tool to make other tests more informative. Think of it as clearing away the noise of pain so the structural signal comes through more clearly.

Arthroscopy as the Gold Standard

When imaging and clinical exams leave genuine uncertainty, or when surgery is already planned for another reason, arthroscopy offers the most definitive answer. A small camera inserted through a tiny incision lets the surgeon visualize the rotator cuff directly, probe the tendon for partial tears that imaging missed, and measure the size and depth of the damage. Because imaging accuracy for partial-thickness tears can be as low as 63% to 75%, arthroscopy remains the reference standard against which all other tests are judged.¹³ The downside, of course, is that arthroscopy is a surgical procedure requiring anesthesia and carries its own risks, so it is reserved for cases where non-invasive methods have not provided a clear picture.

Why a Tear on Imaging Does Not Always Mean Trouble

One of the most important things to understand about rotator cuff diagnosis is that finding a tear on a scan does not automatically explain your pain. Rotator cuff tears are common in people who have no shoulder symptoms at all. A study of asymptomatic volunteers found tear prevalence climbed steeply with age: about 13% in people aged 50 to 59, 20% in people 60 to 69, 31% in people 70 to 79, and a striking 51% in people over 80. The authors described rotator cuff tears as, to some extent, normal degenerative wear rather than a guaranteed source of pain.¹⁹

A more recent systematic review underscored the breadth of this phenomenon. Across population-based samples, full-thickness tears on ultrasound appeared in 11% to 17% of asymptomatic shoulders, and partial-thickness tears or tendon changes showed up even more frequently, reaching as high as 70% in some athlete populations.²⁰ The clinical takeaway is that imaging results always need to be interpreted alongside the exam findings and your symptom pattern. A tear found on MRI in a shoulder that moves well and does not hurt may not need treatment.

Nerve Testing Around the Shoulder

Electromyography, or EMG, occasionally enters the diagnostic picture, but not to detect the rotator cuff tear itself. Large and massive tears can stretch or compress the suprascapular nerve, which supplies the supraspinatus and infraspinatus muscles. When nerve damage is suspected, EMG can detect abnormal electrical activity in those muscles. In a study of patients with massive rotator cuff tears, about 12% showed neurologic findings on EMG, including suprascapular neuropathy and axillary nerve palsy, though fatty infiltration of the muscles did not correlate with nerve conduction changes.²¹

EMG also helps differentiate a rotator cuff problem from cervical radiculopathy or brachial plexopathy, conditions that originate in the neck or nerve plexus but can mimic rotator cuff symptoms with shoulder weakness and pain. One large case series of 659 patients used needle EMG to sort out whether weakness was coming from the rotator cuff, the suprascapular nerve, the axillary nerve, or the cervical spine.²² However, animal model research suggests EMG’s ability to diagnose suprascapular nerve injury caused specifically by a rotator cuff tear is limited, with sensitivity and specificity both falling below 75%.²³ So EMG is more useful for ruling out competing diagnoses than for confirming the tear itself.

Cost and Practical Considerations

If both ultrasound and MRI can detect rotator cuff tears reliably, you might wonder why the choice between them matters. Cost is a real factor. A cost-effectiveness analysis for symptomatic full-thickness supraspinatus tears found that ultrasound was the least expensive strategy at roughly $1,385 per patient, while MRI was the most effective in terms of quality-adjusted outcomes. The incremental cost-effectiveness ratio for MRI came in below the commonly used willingness-to-pay threshold, and in probabilistic modeling, MRI was the preferred strategy about 78% of the time.²⁴ In practice, this means that when the clinical question is narrow and a skilled sonographer is available, ultrasound can be an excellent and cheaper first step. But when the full picture of the shoulder matters, and especially when surgery is being considered, MRI provides information that ultrasound cannot easily match, such as muscle quality and fatty infiltration.

Cost-effectiveness calculations also shift when additional procedures are on the table. Before shoulder replacement surgery, for instance, surgeons need to know whether the rotator cuff is intact. A decision-modeling study evaluated strategies ranging from no advanced imaging to selective or universal MRI or ultrasound, highlighting that the right approach depends on the clinical context and the consequences of a missed tear.²⁵

Artificial Intelligence in Rotator Cuff Imaging

AI tools for reading shoulder MRIs are developing quickly. A systematic review of AI models trained to detect rotator cuff tears on MRI found accuracy figures ranging from 71% to 100%, and statistical comparisons showed no significant differences between AI and human experts in accuracy, sensitivity, or specificity.²⁶ One recently proposed model achieved an average accuracy of about 95% and used visualization techniques to confirm that it was focusing on anatomically relevant regions of the tendon rather than incidental image features.²⁷

These tools are not replacing radiologists yet. Most have been validated only on the data they were trained on, and regulatory approval for clinical use is still in progress for the majority. The likely near-term role is as a second reader, flagging studies that deserve closer attention and potentially reducing the number of tears that slip past a busy radiologist. For patients, the practical implication is that AI-assisted reads may become a standard part of the imaging pipeline within the next several years, potentially improving consistency without changing the scan itself.