Sleeper stretch: this is one of the most recommended exercises for people with a stiff shoulder in internal rotation, especially in sports with the arm overhead. It is also one of the exercises surrounded by the most promises: regaining degrees of movement, preventing rotator cuff injuries, even correcting posture.
This guide does something different from the others: it tells you what the research has really measured. The honest summary is that the exercise has a real but modest effect on mobility, that it is not the best in its category, and that the promise of preventing injuries does not hold up. That is not a reason to throw it away: it is a reason to use it for what it can do, within a programme that also includes other things.
⚠ Important: read this before continuing
The exercises and advice in this article are general indications and may not suit your situation. Every person has a unique clinical history: what works for others could be counterproductive for you.
Want a professional's opinion? Start with the questionnaire: you register and see who will follow you, a physiotherapist registered with the Register chosen for your case, whom you can write to in chat. If you would rather hear a voice, the first consultation is free and we look at your situation together.
When shoulder pain is not "stiffness": the signs not to ignore
Before talking about stretches: the shoulder is one of the sites where the body projects pain that originates elsewhere. Before treating it as a flexibility problem, rule these situations out. If you recognise one of these pictures, do not stretch: get assessed.
- Shoulder pain (especially on the left) together with chest tightness, shortness of breath, cold sweating, nausea or jaw pain. This could be a cardiac picture: the emergency number, straight away.
- Pain at the tip of the shoulder that appeared after an abdominal injury, or together with severe abdominal pain, fever or vomiting. An irritated diaphragm projects pain to the shoulder.
- Fever with a hot, swollen, red shoulder that is extremely painful with any movement. A joint infection must be ruled out: it is an emergency.
- Recent trauma (a fall, an accident, a dislocation) with deformity, or inability to raise or rotate the arm. Imaging first, then exercises.
- Progressive loss of strength, wasting of the shoulder, persistent tingling or numbness in the hand, especially if you are or were a smoker: neurological and chest investigations are needed.
- Night pain that does not depend on position, unexplained weight loss, personal cancer history. These need evaluating before attributing everything to capsular stiffness.
- Pain that steadily worsens over weeks despite relative rest.
Outside these cases, the posterior stiffness of the shoulder is almost always a benign mechanical problem, and that is where the discussion of stretching makes sense. If you are not sure which category you fall into, the guide on when to worry about shoulder pain will help you find your way.
What the sleeper stretch is and where it comes from
The sleeper stretch is a stretch for internal rotation of the shoulder performed lying on your side, hence the name, "the sleeper's stretch". The supporting side fixes the shoulder blade against the surface, and this is the real reason the exercise exists: if the shoulder blade is not still, when you rotate the arm inwards part of the movement is taken by the shoulder blade, and the posterior tissues of the shoulder are not stretched at all.
The target is the posterior structures of the glenohumeral joint: the posterior capsule, the rear part of the deltoid, and the infraspinatus and teres minor muscles. When these tissues are stiff one speaks of posterior shoulder tightness, and when the consequence is reduced internal rotation compared with the other side one speaks of GIRD, the English acronym for glenohumeral internal rotation deficit.
Here comes the first important correction to how this topic is usually presented. GIRD is not all soft tissue. In the dominant arm of someone who has thrown for years, part of the internal rotation "lost" is a bone adaptation: the humerus has remodelled by rotating backwards, a phenomenon called humeral retroversion. That portion does not stretch, because it is not elastic tissue: it is bone geometry. Another part is due to the transient properties of the muscle after exertion, which resolve on their own. Only the remaining portion, the capsulo-muscular one, is what a stretch can realistically influence. The literature in fact distinguishes an anatomical, adaptive GIRD that should be left alone from a pathological GIRD. Measuring the difference between the two sides with a goniometer and treating all of it as "stiffness to stretch" is the conceptual error underlying many of the programmes found online.
What the research really says
This is the part that is usually missing. It is worth reading in full, because it changes the way you will use the exercise.
On mobility: the effect exists, but is modest and not exclusive
A systematic review of randomised trials on stretching for posterior stiffness and GIRD found moderate-quality evidence in favour of the cross-body stretch, the stretch in which the arm is brought across the front of the chest, with immediate and short-term effects in young, asymptomatic subjects. On the sleeper stretch the same review reached an uncomfortable conclusion: moderate evidence suggests that the active version of the exercise may not be more effective than doing nothing, again in the short term.
It is worth reading that sentence carefully, because it is easy to misunderstand in both directions. It does not say the sleeper stretch is useless: it says that in the version where you actively push yourself, on the parameters and timescales studied, it did not stand out from the control.
A second synthesis reaches a more favourable reading, but it needs to be read knowing what it pulled together. It is a scoping review, the type of review that gathers all the available literature, including non-randomised studies, without weighting their quality as a systematic review would, and it selected 21 articles out of 2,777. Its conclusion is that cross-body and sleeper stretches are the two most studied treatments and appear effective in reducing posterior stiffness, with a practical detail that is not trivial: stabilising the shoulder blade during the stretch optimises the stretch and the gains in horizontal adduction.
For that review, however, there is a warning. Its stated aim was to build a treatment framework that could be both taught and delivered by the therapist, and among the interventions it surveyed are also the cross-body and sleeper stretches performed by the physiotherapist on the patient, together with instrumental soft-tissue mobilisation, muscle energy techniques, dry needling and assisted stretching protocols. The favourable conclusion therefore does not concern only the exercise you do alone on your living-room mat: it also covers the versions in which a professional positions and loads the shoulder. This is not a reason to ignore it: it is a reason not to read it as proof that doing it yourself is enough.
The overall picture, then, is neither "it works" nor "it does not work". It is this: the effect on mobility exists, it is of limited size, and it is not a special property of the sleeper stretch. A simpler alternative stretch achieves at least as much.
On pain: the best contribution comes within a programme
On people who really have pain the evidence is more encouraging, and it is useful to know how it was obtained. A randomised study assigned 67 patients with subacromial impingement syndrome and internal rotation asymmetry to three groups: a treatment programme plus a modified cross-body stretch (22 people), a treatment programme plus a modified sleeper stretch (22 people), or the treatment programme alone, consisting of physical therapies, mobility and strengthening, without posterior stretches (23 people), for four weeks.
All three groups improved. But the two groups that had added a posterior stretch did better than the comparison group on pain during activity, internal rotation, function and disability, with improvements judged clinically significant. Between the two stretches, once again, there was no difference.
Two things follow from this. The first: the added value measured is relative to a complete programme that already contained strengthening and mobility, not relative to nothing: the stretch is the extra ingredient, not the recipe. The second: the stretches used in that study were the modified versions, not the classic ones. This is not a detail, and it is the reason I describe the technique in that way further on.
On preventing injuries: here the promise does not hold
This is the claim read most often and the one the evidence supports least. The usual chain of logic is: rigid posterior capsule → abnormal stress on the tendons → rotator cuff tear; therefore stretching prevents it. Every link in that chain is open to question.
A 2020 review devoted precisely to this question concludes that GIRD on its own is not associated with the risk of upper limb injury in the overhead athlete. Other variables are: external rotation deficit, horizontal adduction deficit and shoulder flexion deficit have been associated with injuries, whereas GIRD as such is not. If the target is not the risk factor, hitting it is not automatically a prevention strategy.
The same source, however, adds two qualifications that it would be dishonest to leave out. The first: the earlier literature had instead associated pathological GIRD with an increased injury risk: it is the recent data that have scaled back the association, not a consensus that has always existed. The second is more insidious: the authors hypothesise that the association has weakened precisely because, in recent years, pathological GIRD is recognised and treated early. If that were so, the absence of risk would not prove the problem does not matter; it would prove that treating it works. This remains a hypothesis, not a result, and it is the reason why the correct formula here is "not proven", not "disproven".
There is also an experimental finding pointing in the same direction. A randomised study of high school baseball pitchers compared, every evening for a 150-day monitoring period, a stretching programme (62 participants, active control group) against a programme of external rotator strengthening (51 participants). Shoulder and elbow injuries were 22.6% (14 athletes) in the stretching group and 9.8% (5 athletes) in the strengthening group, with a lower risk for those who strengthened (hazard ratio 0.489). The study was designed to show that strengthening was not inferior to stretching, and the result went beyond that goal.
Read it carefully, for two reasons. The first: there was no group that did nothing, so the study does not show that stretching is useless. It shows that, with the same evening commitment, the boys who strengthened the external rotators were injured less than those who did the sleeper stretch. If you have to decide where to put your ten minutes, that finding matters.
The second reason is uncomfortable for the thesis of this section, and passing over it would be wrong: the authors of that same study start from the opposite premise. In their introduction they write that internal rotation deficit and weakness in external rotation are risk factors for shoulder and elbow injury in high school pitchers, and that a stretching programme to reduce GIRD lowers the injury rate. The two readings do not reconcile on their own: the 2020 review looks at overhead athletes in general and finds that isolated GIRD does not predict injury; this research group works on a single, highly exposed population, where it considers it a risk factor. For this reason the correct formula remains "not proven" rather than "refuted", with the addition that if you pitch at competitive level, you are in the population where the preventive hypothesis is less weak, not more so.
And posture?
In the most widespread version of this topic you read that posterior stretching rebalances tensions and corrects rounded shoulders. I have found no evidence to support this claim, and I have therefore not included it in the article. If someone sells it to you as a reason to do the exercise, they are extrapolating.
An important tip: before trying these exercises, make sure you have had a correct diagnosis. People often think they have one problem, but the real cause of the pain is something else. Doing the wrong exercises can make things worse rather than better.
How to do it: the modified version
The classic sleeper stretch has a recognised problem: lying on your side with the shoulder flexed to 90 degrees, as you rotate the arm internally you compress the subacromial space against the weight of your own body. For this reason, the potential for subacromial conflict with the sleeper stretch, and the inability to control the shoulder blade and rotation with the classic cross-body, a group of physiotherapists who work with throwers proposed a modification of both exercises. It is a clinical commentary, meaning a reasoned opinion and not an experiment; but it is the version that later clinical studies adopted, and it is therefore the one for which patient data exist.
Here is how to set it up:
- Lie on your side, on the side of the shoulder to be treated, knees bent for stability.
- Rotate your trunk slightly backwards, just enough to take the weight of your body off the shoulder: you do not need to stay perfectly perpendicular to the surface. You can place a cushion or a rolled towel behind your back to hold the position. This rotation is the key modification: it removes the direct load on the upper part of the shoulder.
- Bring the arm forward under you, elbow bent, in a position slightly below the line of the shoulder rather than exactly at 90 degrees. If you feel tingling in the upper or front of the shoulder, you are too high: lower the arm.
- Make sure the shoulder blade stays in contact with the surface. This is the only thing that makes the exercise different from any ordinary movement. If the shoulder rolls forward, lifting off the surface, your stretch has ended there.
- With your other hand, guide the forearm towards the surface, without pushing. The correct sensation is a dull pull behind the shoulder, not a sharp pain, not a pinch, not tingling.
- Breathe normally and hold until the tension softens, then slowly return.
On the exact duration of the hold, the number of repetitions and the optimal frequency, the literature does not offer a proven protocol superior to the others, and the studies cited here are not even comparable with each other: in the study on subacromial conflict pain the programme lasted four weeks, in the study on pitchers the exercises were done every evening and monitoring lasted 150 days. Regulate yourself by sensation, not by a timer found online.
Mistakes that cancel out the exercise
- Pushing hard. Capsular stiffness does not give way to force: if the shoulder reacts by tensing, you have achieved the opposite. If acute pain, pinching or tingling appear, reduce immediately.
- Letting the shoulder blade lift off the surface. It is the mistake that makes the exercise useless without your noticing: you go on feeling a movement, but it is the shoulder blade that is sliding, not the glenohumeral joint that is rotating.
- Staying perfectly on your side. It is the variation that compresses the subacromial structures most. Rotate the trunk backwards.
- Keeping the arm too high. Above the line of the shoulder the conflict is more likely. Lower it a few degrees.
- Holding your breath. It increases general muscle tone and works against what you are trying to do.
- Pushing through a shoulder that hurts. If today's session leaves more discomfort than yesterday's, it is not "the tissue freeing up": it is a signal.
Who it makes sense for, and who less so
It is worth assessing if you practise a sport with the arm overhead, such as tennis, volleyball, swimming, baseball or handball, and have a difference in internal rotation between your dominant arm and the other, especially if accompanied by posterior discomfort. Those who play volleyball will find a more specific picture in the guide on shoulder injury prevention for volleyball players.
It makes sense within a pathway if you have subacromial conflict pain that has already been assessed by a professional: this is the population on which the best evidence has been collected, and always in addition to a programme, never on its own. If this is your case, the article on shoulder impingement explains the full picture.
It makes less sense than is often claimed in two situations. The first: people who work at a computer. It is a plausible association but not a proven one, and desk-related shoulder pain usually has other origins: this is discussed in the guide on mouse shoulder pain from computer use. The second: those who do it as pure prevention without either symptoms or asymmetry. In that case, in light of the data above, time is better spent strengthening.
Contraindications and precautions
Avoid the exercise, and get assessed first, if you have had a dislocation or shoulder instability: forced internal rotation of a joint that already moves too much is not what is needed. The same applies to a recent injury or surgery without the indication of your surgeon or physiotherapist.
If you have subacromial conflict syndrome, the exercise may be appropriate, but in the modified version and within a programme, not in the classic version done on your own initiative. The risk of the classic version is precisely the compression of structures that are already irritated.
Stop immediately and seek an assessment if during the exercise pain radiates down the arm, tingling in the fingers or numbness appear: these are symptoms of nerve origin, not of capsular stiffness.
The cross-body stretch: try it first
If you have read this far, the practical conclusion is fairly clear. The cross-body stretch, with the arm brought across the front of the chest and supported with the other hand, has evidence of effectiveness at least equal to the sleeper stretch in some of the better syntheses, and offers three concrete advantages: it can be done standing or seated, it does not compress the shoulder against a surface, and it is easier to perform without getting it wrong.
Here too, however, the shoulder blade is everything. In the modified version you press the forearm of the other arm against the elbow, keeping the elbow bent, while holding the shoulder blade still against a wall or against the floor if you are lying on your back. Simply bringing the straight arm across the body, as is seen done in the gym before training, mainly produces shoulder blade sliding.
For most people the reasonable choice is to start there, and to consider the sleeper stretch as an alternative if the cross-body feels uncomfortable or ineffective on your body.
What to put alongside it
Given what we know, a sensible programme for an overhead shoulder looks more like a strength programme with some mobility inside it than a stretching session.
Strengthening the external rotators is the strongest candidate in light of the randomised study cited above, and it is the part that is missing altogether from most home programmes. The rotator cuff exercises cover this part, while the general shoulder strengthening exercises serve for the shoulder girdle as a whole.
It also makes sense to work on horizontal adduction, the movement of the arm across the chest, and on external rotation, because these are the deficits that in the literature are associated with injuries, unlike isolated GIRD. These are precisely the parameters that a stretch aimed only at internal rotation does not touch.
If nothing changes after a few weeks
If you practise consistently and mobility does not move, the most likely explanation is not that you are doing the repetitions wrong. There are two possibilities, and neither is resolved by pushing harder.
The first: the missing range is bony. If the internal rotation you have lost depends on humeral retroversion, no amount of stretching will restore it, and it is an adaptation, not a defect. Continuing to force a shoulder that has already given up everything it could only load the anterior capsule.
The second: the problem was not stiffness. Rotator cuff tears, labral changes, glenohumeral arthritis, frozen shoulder, cervical causes of referred pain. These are pictures that behave differently and need recognising, not stretching.
In both cases the next step is an assessment, not another month of sleeper stretch. Through telerehabilitation a physiotherapist can watch you perform the movement, see whether the shoulder blade really stays still, measure the difference between the two sides and understand whether it makes sense to continue or change course. Ri-Hub's online physiotherapy sessions cost 29,90€ and are done from home: for a technique problem and the choice of exercise, this is often enough.
In summary
The sleeper stretch is a reasonable exercise with a reputation larger than its results. On mobility the effect exists but is limited, and the cross-body stretch matches it with better evidence and less risk of irritating the shoulder. On pain it makes a real contribution, but within a complete programme and in the modified version. On injury prevention the promise does not hold as it is sold: according to the most recent review on this question, internal rotation deficit alone is not shown to be associated with risk, on this, however, the literature is divided, and in the direct comparison among pitchers, external rotator strengthening did better.
The conclusion is not "do not do it". It is: do it well, with the shoulder blade still and without forcing, knowing that it is a small piece of a programme that must contain mostly strength, and knowing how to recognise when the problem you have is not stiffness.
Sources
- Effectiveness of Stretching on Posterior Shoulder Tightness and Glenohumeral Internal-Rotation Deficit: A Systematic Review of Randomized Controlled Trials, Journal of Sport Rehabilitation, 2017
- Glenohumeral Internal Rotation Deficit: Prime Suspect or Innocent Bystander?, Current Reviews in Musculoskeletal Medicine, 2020
- Effects of Modified Posterior Shoulder Stretching Exercises on Shoulder Mobility, Pain, and Dysfunction in Patients With Subacromial Impingement Syndrome, Sports Health, 2020
- Shoulder stretching versus shoulder muscle strength training for the prevention of baseball-related arm injuries: a randomized, active-controlled, open-label, non-inferiority study, Scientific Reports, 2022
- An evidence-informed rehabilitation management framework for posterior shoulder tightness: A scoping review, Shoulder & Elbow, 2024
- The modified sleeper stretch and modified cross-body stretch to increase shoulder internal rotation range of motion in the overhead throwing athlete, Journal of Orthopaedic & Sports Physical Therapy, 2013
Last source check: August 2026.
You might also be interested in:
Epicondylitis: Causes, Symptoms and Treatment of Tennis Elbow Back Stretching Exercises: Complete Routine to Lengthen and Loosen Walking Exercises: How to Turn a Stroll into a Workout
Stay in touch with us Join our dedicated Facebook group: we answer all patients, give practical advice and clear up the most common questions.
Need a consultation?
Start with the questionnaire on Ri-Hub: a few questions about your case, then you register and see who will follow you: a physiotherapist registered with the Register, whom you can write to in chat whenever you like. If you would rather talk it through by voice, the first consultation is free, lasts twenty minutes and commits you to nothing: it starts with them already having read what you wrote, and together you decide whether and how to go on.
