Full Can Test vs Empty Can Test for the Supraspinatus

The full can test and the empty can test both aim to evaluate the supraspinatus muscle and tendon, but they do it in meaningfully different arm positions, and neither one is clearly superior across the board. Both are variations of what clinicians call the Jobe test, and research consistently shows they land at similar overall diagnostic accuracy for detecting supraspinatus tears. Where they diverge is in how much pain they provoke, how selectively they load the supraspinatus versus surrounding muscles, and how reliably different examiners agree on the result. Those differences matter more in practice than most textbooks let on.

How the Two Tests Work

Both tests start from the same basic setup: your arms are raised to about 90 degrees in the scapular plane, roughly 30 to 45 degrees forward of straight out to the side. The examiner pushes down on your arms while you resist. The difference is what your thumbs are doing.

In the empty can test (sometimes called the Jobe test in its original form), you internally rotate your arms so your thumbs point toward the floor, as if pouring out a can. In the full can test, you externally rotate so your thumbs point toward the ceiling, as if holding a full drink you don’t want to spill. The examiner is looking for two things: whether you feel pain, and whether one side is weaker than the other.

That rotation difference seems small, but it changes which muscles are working hardest, how the shoulder blade moves, and how the tendons sit under the bony arch above them. Understanding those differences is the key to knowing when each test is more useful.

What the Tests Actually Activate

A persistent belief in clinical teaching is that the empty can position “isolates” the supraspinatus. The evidence tells a more complicated story. An EMG study recording 13 shoulder muscles in healthy subjects found that during the empty can test, eight or nine other muscles were activated at levels comparable to the supraspinatus. The full can test was similar, with eight muscles matching supraspinatus activity. The authors concluded that neither test satisfies basic criteria for selectively activating the supraspinatus with minimal contribution from other muscles.¹

That said, the empty can position does appear to push the supraspinatus harder. A study using PET/CT imaging, which measures actual metabolic activity in muscles during exertion, found that the supraspinatus showed significantly higher activity during the empty can test than the full can test. The middle deltoid and subscapularis also ramped up more in the empty can position.² Another EMG study confirmed this pattern, finding greater average activation of both the supraspinatus and subscapularis during the empty can test.³

So the empty can test demands more from the supraspinatus, but it also demands more from nearly everything else. It is a harder test overall, not a more selective one. This distinction matters clinically because if a patient fails the empty can test, you cannot confidently attribute that failure to the supraspinatus alone.

Diagnostic Accuracy for Supraspinatus Tears

When you compare how well each test identifies actual tears confirmed on imaging, the results tend to converge. One prospective study found the empty can test had a sensitivity of about 89% and specificity of 59% for supraspinatus tears (both full- and partial-thickness), while the full can test came in at 75% sensitivity and 47% specificity.⁴ On those numbers alone, the empty can test looks better at catching tears. But a different study found the full can test reached a sensitivity of 70% with a specificity of 81%, a stronger specificity figure than most empty can data.⁵

A more recent comparison using MRI-confirmed tears reported that both tests landed at identical overall diagnostic accuracy of about 77%. The empty can test was more sensitive (82% versus 74%), while the full can test was more specific (81% versus 67%).⁶ This trade-off is consistent with the broader pattern: the empty can test casts a wider net and catches more tears but also produces more false positives. The full can test misses a few more tears but is better at correctly ruling out people who don’t have one.

Neither test performs well enough alone to confirm or exclude a supraspinatus tear. A study of the Jobe test in both single-arm and double-arm formats found accuracy values ranging from roughly 46% to 60%, with sensitivity varying widely depending on whether pain, weakness, or both were used as the positive criterion.⁷ These are not numbers that let you skip an MRI.

The Pain Problem

Pain during testing is a double-edged sword. On one hand, provoking pain can be diagnostically useful. On the other, pain inhibits muscle activation and can make a patient look weak even when their tendon is intact, muddying the strength-based interpretation the examiner is relying on.

The empty can position consistently hurts more. In one of the earliest head-to-head comparisons, pain was observed in about 50% of shoulders during the empty can test versus 43% during the full can test, though that gap was not statistically significant. The authors noted that because pain interferes with accurate strength testing, the full can test’s lower pain provocation could be an advantage in the clinic.⁸ The more recent MRI-confirmed study found a larger and statistically significant gap, with average pain scores of about 4.1 on a 10-point scale during the full can test versus 5.9 during the empty can test.⁶

Why does the empty can position hurt more? Internal rotation narrows the space under the acromion and mechanically loads the supraspinatus tendon in a way that compresses it against bony structures. For someone already dealing with tendon irritation or a partial tear, that compression translates directly into pain. The full can position opens things up slightly with external rotation, reducing the pinch on already irritated tissue.

This pain difference creates a real clinical dilemma. If a patient grimaces and gives way during the empty can test, was that weakness from a torn tendon or pain-driven inhibition from a perfectly intact but inflamed tendon? Research looking at this question found that pain during testing explained about 36% of the variance in observed weakness, essentially accounting for most of the strength deficit on its own.⁹ In other words, pain makes people test weak regardless of whether they have a structural tear.

Weakness Versus Pain as a Diagnostic Criterion

This brings up a practical question that doesn’t get enough attention: should you interpret the test based on weakness, pain, or both? The answer depends on what you’re trying to differentiate.

For detecting a structural supraspinatus tear, weakness is the more useful finding. The original Itoi study from 1999 was explicit about this: muscle weakness should be the primary indicator of a torn tendon, and on that basis both tests perform equivalently.⁸ But weakness alone has a pitfall. A study comparing rotator cuff tears to cervical spondylotic amyotrophy, a nerve-related condition that can mimic cuff weakness, found that weakness during both tests could not distinguish between the two diagnoses because of low specificity. Pain provocation, however, was useful for telling them apart because it was much more specific to the rotator cuff condition.¹⁰

The takeaway is that a clinician ideally notes both findings separately rather than lumping them together as a single positive or negative result. Weakness points toward a structural tear, while pain points toward a rotator cuff origin of the problem but doesn’t specify whether the tendon is torn or just irritated.

Shoulder Blade Mechanics and Subacromial Space

One argument sometimes made for the full can test is that the empty can position narrows the subacromial space, the gap between the humeral head and the bony arch above it where the supraspinatus tendon runs. If that were true, the empty can position would be inherently more provocative and less reliable for strength testing. The biomechanical data here is surprisingly nuanced.

A study directly measuring acromiohumeral distance in both positions found no difference between the empty can and full can arm positions, and no difference between patients with shoulder pain and pain-free controls.¹¹ The two positions did produce different scapular kinematics, though. The empty can position caused more scapular upward rotation and less posterior tilt, while the full can test showed the opposite pattern. But these kinematic differences did not translate into a measurable change in the actual space available for the tendon.

Interestingly, a separate study investigating why the full can test provokes pain in patients with subacromial pain syndrome found that a positive full can test was associated with a reduction in acromiohumeral distance and an increase in scapular anterior tilt. The authors proposed that these mechanical changes could lead to tendon compression, though they couldn’t confirm direct compression with their methods.¹² This suggests that both positions can produce impingement-like mechanics depending on the patient’s particular shoulder movement patterns.

Examiner Agreement and Reproducibility

A test is only useful if different clinicians get the same result when they perform it on the same patient. The full can test appears to have an edge here. The MRI-confirmed comparison study found interobserver reliability was higher for the full can test, with a kappa statistic of 0.80 (considered substantial agreement) compared to 0.63 for the empty can test (moderate agreement).⁶

The likely explanation ties back to pain. When patients experience more pain, their effort and guarding become more variable, making the examiner’s judgment call harder. The full can position’s lower pain provocation produces a cleaner strength signal, making it easier for two examiners to agree on whether the patient is genuinely weak. This is an underrated practical advantage. In busy clinical settings where different therapists or residents might examine the same patient at different visits, reliability matters as much as raw diagnostic accuracy.

Angle Variations and Testing Modifications

The debate doesn’t stop at internal versus external rotation. The angle of arm elevation also changes what the test picks up. A study comparing the empty can test performed at 0 degrees of scaption (arm at the side) versus 30 degrees found dramatically different sensitivity-specificity profiles. At 0 degrees, sensitivity for full-thickness tears was about 89% but specificity dropped to roughly 25%. At 30 degrees, the pattern reversed: sensitivity fell to about 39% while specificity climbed to 63%.¹³ This means the standard testing angle of 90 degrees is only one option, and clinicians who modify the elevation angle are essentially trading sensitivity for specificity depending on what they need from the test in that moment.

When it comes to whether the supraspinatus is working equally hard across positions, ultrasound measurement of muscle cross-sectional area during isometric contraction found no significant differences between the empty can, full can, and a diagonal horizontal adduction test. All three showed significantly greater cross-sectional area compared to the resting state, confirming they all activate the muscle, but none appeared to load it meaningfully more than the others.¹⁴

The Champagne Toast Alternative

Given the limitations of both traditional positions, some researchers have looked for better ways to isolate the supraspinatus. One promising candidate is the “champagne toast” position: arm at about 30 degrees of abduction, mild external rotation, 30 degrees of forward flexion, with the elbow bent to 90 degrees, as if you were raising a glass. An EMG study found this position produced a supraspinatus-to-deltoid activation ratio of about 4.6, compared to just 0.8 in the traditional Jobe (empty can) position.¹⁵

That ratio difference is striking. A ratio below 1.0 means the deltoid is actually working harder than the supraspinatus during the Jobe test, which fundamentally undermines the idea that the empty can position targets the supraspinatus. The champagne toast position flips this so the supraspinatus is doing roughly four to five times more work than the deltoid. The lower elevation angle also keeps the arm well below the impingement zone, which should reduce pain provocation. This test hasn’t yet been widely validated for diagnostic accuracy in large clinical studies, but the biomechanical rationale is strong enough that it’s gaining traction in sports medicine and orthopedic rehabilitation settings.

When Weakness Doesn’t Mean a Tear

One misconception worth addressing is the assumption that a positive empty or full can test means you have a supraspinatus tear. These tests can be positive for a range of reasons: tendinopathy without a tear, bursitis, impingement syndrome, cervical nerve root problems, or even simple deconditioning. As noted earlier, weakness during both tests failed to distinguish rotator cuff tears from cervical spondylotic amyotrophy.¹⁰ This means that a positive test based on weakness alone should prompt further investigation rather than a definitive diagnosis.

The supraspinatus also doesn’t work in isolation in the living shoulder. Cadaver studies have shown that when the supraspinatus is disabled, the deltoid can compensate by increasing its force output by about a third of the lost supraspinatus contribution, recovering most of the lost elevation.¹⁶ In real patients with chronic tears, the deltoid and remaining rotator cuff muscles often adapt substantially over time, which is why some people with large full-thickness tears still pass these tests. Computer modeling supports this: without the supraspinatus, other muscles increase their force output by roughly 30% to compensate, though this comes at the cost of increased upward migration of the humeral head and greater eccentric loading on the joint.¹⁷ A false-negative result on either test, where the patient appears strong despite having a tear, is therefore not a testing error so much as a reflection of how the shoulder adapts.

Choosing Between Them in Practice

If you’re a clinician deciding which to use, the honest answer is that using both, along with other tests in a cluster, gives you the best picture. But if you have to pick one, the choice depends on what you’re optimizing for. The empty can test is slightly more sensitive and will catch a few more tears, making it better as a screening tool when your goal is not to miss anything. The full can test is more specific, less painful, and more reproducible between examiners, making it a better choice when you want a cleaner strength assessment or when the patient is in too much pain to give a reliable effort in the internally rotated position.

For patients who are acutely painful, starting with the full can test is pragmatic. If they can tolerate it and show clear weakness, you have useful information without needing the more provocative position. If the full can test is negative but clinical suspicion remains high, adding the empty can test or the champagne toast position can increase your diagnostic yield. In rehabilitation settings where you’re tracking strength recovery over time, the full can test’s higher interobserver reliability makes it the more practical choice for serial assessments where you want to compare results across visits.

Neither test replaces imaging when a definitive structural diagnosis is needed. Their real clinical value is in guiding decisions about who needs imaging and how urgently, not in making the final call about what’s happening inside the shoulder.