Rotator cuff exercises: this is what people search for when they have pain in the shoulder and want to understand what they can do on their own. The honest answer is that it depends on what you have: the same band is the right treatment for a tendinopathy and wasted time, sometimes precious time, for a complete traumatic tear.
This guide separates the clinical pictures instead of treating them as one, reports the real numbers from studies (even when they are less exciting than you might expect) and explains when the programme should be built with a professional rather than copied from an article.
⚠ Important: read this first
The exercises and advice in this article are general guidelines and may not be suitable for your situation. Every person has a unique clinical history: what works for others could be counterproductive for you.
Thinking about starting a course? Answer a few questions: you register and see who will follow you, a physiotherapist registered with the Albo chosen for your case, whom you can write to in chat. If you would rather hear their voice, the first consultation is free and we look at your situation together.
First of all: the signs that call for an assessment, not a band
There are situations in which exercises are not the first move. If you recognise yourself in one of these, get assessed instead of starting a programme:
A separate case, often mistaken for a cuff problem: if pain and stiffness have taken hold of both shoulders together, appeared within a few weeks after the age of 50, and last more than an hour in the morning, the suspicion shifts to an inflammatory cause, see polymyalgia rheumatica. There you need a doctor, not an exercise programme.
- Recent trauma followed by loss of strength: you fell, took a knock or tore something, and now you cannot lift the arm or only lift it with help from the other hand. This is the picture that raises suspicion of an acute full-thickness tear.
- Trauma with suspected dislocation: a deformed shoulder, blocked movement, a feeling that the joint has "come out".
- Night pain that gets worse day by day, instead of fluctuating.
- Fever, or swelling or redness of the shoulder.
- Tingling or widespread weakness of the whole arm and hand.
To put it plainly: if the tendon has completely detached after a trauma, exercises alone will not reattach it. They can maintain mobility and reduce pain, but they do not restore the continuity of the tendon, and waiting has a cost. In a retrospective series of 206 traumatic full-thickness tears that were later surgically repaired, those operated on within four months of the injury had better functional scores than those who came later (+10.3 ASES points), with the best results in the few operated on within three weeks (Gutman et al., 2021). It is a retrospective, non-randomised study, and it compares timings among people who were all operated on: it does not show that surgery beats rehabilitation. It does say, however, that in that picture the months spent trying alone are not neutral.
If any of these signs apply to you, the urgent symptoms page explains where to go and how quickly.
The four muscles and what each one does
The rotator cuff is made up of four muscles whose tendons wrap around the head of the humerus like a cuff. They are not powerful motors: their job is to keep the humeral head centred while the large muscles (deltoid, pectoralis major, latissimus dorsi) move the arm.
- Supraspinatus: initiates abduction, that is the lateral lifting of the arm. It is the tendon that tears most often and the one involved in the great majority of the pictures we discuss here.
- Infraspinatus: the main external rotator.
- Teres minor: works with the infraspinatus in external rotation and in posterior control.
- Subscapularis: the main internal rotator and the front part of the cuff. When it is involved in a tear, the picture is generally more serious and the conclusions of studies on "small supraspinatus tears" no longer apply.
Understanding this division serves a practical purpose: a programme that trains only external rotation leaves out half of the cuff, and one that ignores the scapula leaves out the base on which the cuff works.
Four different pictures, four different strategies
This is the distinction that is almost always missing from generic articles, and it is the one that changes what you should do today.
1. Tendinopathy and subacromial pain (the "impingement"). Pain that appears with movements above the head, often without a specific trauma, with strength essentially preserved. This is the picture in which progressive exercise is the first-line treatment, and it is the one that most of the recommendations you come across refer to.
2. Partial tear or small degenerative tear. The tendon is frayed or interrupted only in part, usually from wear rather than an event. Here rehabilitation remains a reasonable first choice: in a randomised trial on small, non-traumatic tears of the supraspinatus alone in people over 55, after an average of 6.2 years there were no significant differences between physiotherapy alone, acromioplasty and tendon repair (Kukkonen et al., 2021). The other side also needs stating: in that study, 8 patients in the physiotherapy group had nonetheless gone on to surgery during follow-up.
3. Complete traumatic tear. An identifiable trauma, often acute pain followed by a clear weakness. It is not the same condition as the previous points and should not be managed by the same criteria: here specialist assessment comes before the programme. The point is so concrete that the Cochrane review on rotator cuff surgery explicitly states that its conclusions do not apply to traumatic tears, to large tears involving the subscapularis, or to young patients (Karjalainen et al., 2019).
4. Operated shoulder. After a repair, the sutured tendon goes through phases of biological healing with precise timings. The programme is dictated by the surgeon who operated, together with the physiotherapist: it is not do-it-yourself material, and we come back to this later.
What the studies really say (numbers, not slogans)
Exercise is the first line for non-traumatic cuff pain. But "first line" does not mean "large, guaranteed effect", and it is worth knowing this before you start, because it avoids both inflated expectations and giving up in the first difficult month.
The effect measured against placebo is modest. The Cochrane review on manual therapy and exercise for rotator cuff disease included 60 studies, but only one high-quality study compared a realistic programme of manual therapy and exercise with a placebo. At 22 weeks the difference in pain was 6.8 points out of 100 (confidence interval from -0.7 to +14.3, so compatible with no benefit at all) and in function 7.1 points out of 100. The authors conclude that there are no clinically important differences between the groups (Page et al., 2016).
Doing a lot of exercises is not automatically better. The GRASP trial randomised 708 patients with cuff disorders to a progressive exercise programme (up to 6 physiotherapy sessions) or to a single session of good-practice advice. At 12 months there was no difference in shoulder pain and disability scores (mean difference -0.66; 99% CI from -4.52 to 3.20). In the same study, corticosteroid injection gave no long-term benefit (Hopewell et al., 2021). Note a detail that is often omitted: GRASP excluded people who had a significant shoulder trauma (dislocation, fracture, full-thickness tear needing surgery), frozen shoulder and instability. Those results do not speak to those cases.
In the same direction, a smaller randomised study compared a self-managed programme based on a single loaded exercise with usual physiotherapy for rotator cuff tendinopathy, finding no differences at 3, 6 and 12 months (Littlewood et al., 2016). It was a small trial with many dropouts: it does not prove that the two approaches are equivalent, but it failed to show that the richer programme is superior.
Surgery does not deliver dramatic results for degenerative tears either. The Cochrane review of 9 randomised trials and 1,007 participants found that, at 12 months, repairing the cuff compared with non-surgical treatment improves pain by 0.87 points on a 0-10 scale (moderate certainty) and function by 6 points out of 100 (low certainty): small differences, in studies that mostly included degenerative tears of the supraspinatus (Karjalainen et al., 2019).
But not all studies point in the same direction, and that should be said. A Norwegian trial of 103 patients with small and medium tears (up to 3 cm) found that at 10 years tendon repair was superior to physiotherapy alone, with an advantage that widened over time: +9.6 Constant points, +15.7 ASES and 1.8 cm on the 10 cm pain scale (Moosmayer et al., 2019). In that study, 14 patients in the physiotherapy group went on to surgery as a second step, with results worse than those who were operated on straight away.
The honest conclusion is this: for small, degenerative tears, in people over 55, starting with rehabilitation is a reasonable and supported choice. For traumatic or large tears, or in young and active people, the evidence in favour of conservative care simply is not there, and anyone who sells it to you as a certainty is extending studies to populations those studies had excluded. This is exactly why an assessment is worth more than a generic protocol.
Why an MRI showing a tear does not decide on its own
A finding that surprises many people: cuff tears are common even in people with no symptoms at all.
In a population screening of 664 residents of a Japanese village, 22.1% had a full-thickness cuff tear. Prevalence rose with age, 0% between 20 and 40, 10.7% in people in their fifties, 26.5% in their seventies, 36.6% in their eighties, and 65.3% of those tears were completely asymptomatic (Minagawa et al., 2013).
A very similar result comes from a British cohort of 1,003 women aged 64 to 87: 22.2% full-thickness tears, of which 48.4% without symptoms. In the same study the severity of symptoms was not linked to the severity of the finding, except for tears over 2.5 cm (Hinsley et al., 2022).
Two practical consequences. The first: after 60, a tear found on ultrasound or MRI may be an incidental finding, present for years and not responsible for today's pain. The second, opposite and just as important: if you are under 50, where the baseline prevalence is close to zero, a full-thickness tear that appears after a trauma weighs far more. The image should be read together with the history and the clinical examination, not instead of them.
Phase 1: calming symptoms and keeping movement
This phase is for when you have active pain. The aim is not to strengthen, it is to keep the shoulder moving and get it used to producing force again without triggering a flare-up.
Codman's pendulum. Bent forward, supported by your healthy hand on a table, let the painful arm hang and swing it in small circles, letting gravity do the work rather than the shoulder muscles. 1-2 minutes.
Isometrics in external and internal rotation. Elbow at your side bent to 90°, press the forearm against a door frame or against your other hand without moving anything. Hold for 20-30 seconds at a submaximal force (about 50-70% of maximum), 5 repetitions. It is the way to load the tendon when movement against resistance is still too much.
Posterior capsule stretch (cross-body). Bring the affected arm across your chest and guide it towards you with your other hand. 30 seconds, 3 times.
Doorway stretch. Standing in a doorway, place your forearms on the frame with elbows at 90° and step forward to open the chest. 30 seconds. It stretches the pectorals, which often limit shoulder mechanics.
Sleeper stretch. Lying on the side of the arm to be stretched, elbow at 90°, gently push the forearm towards the bed with your other hand. 30 seconds, 3 repetitions. It improves internal rotation when it is limited; if it flares up the pain, stop it: it is not a mandatory exercise.
Phase 2: strengthening with a band
When resting pain has subsided and isometrics no longer bother you, you move on to resistance. The band is the ideal tool because the resistance increases gradually along the movement.
External rotation with elbow at your side. Band anchored at elbow height, elbow bent to 90° and tucked against your side, rotate the forearm outwards. A rolled towel between elbow and side helps you avoid cheating. 15 repetitions, 3 sets. It works the infraspinatus and teres minor.
Internal rotation with a band. Same position, band anchored on the opposite side, bring the forearm towards your abdomen. 15 repetitions, 3 sets. It works the subscapularis, the part of the cuff that almost everyone forgets.
Scaption (abduction in the plane of the scapula). Standing on the band, raise the arm to about 45°, between forward and sideways, thumb pointing up, up to shoulder height. 15 repetitions, 3 sets. The scapular plane is less provocative than pure lateral lifting.
External rotation at 90° of abduction: advanced phase only. Arm abducted to 90° and elbow at 90°, rotate the forearm upwards against the band. It is the functional position of someone throwing or smashing, but it is also the most provocative: it makes sense later in the pathway, and makes little sense if movements above the head still hurt.
Phase 3: progressive loading and scapular control
Strengthening the cuff without scapular control is half the work: the scapula is the mobile base on which the shoulder moves, and if it is not stable the cuff works at a disadvantage.
Side-lying external rotation. Light weight (0.5-2 kg), elbow at 90° resting at your side, rotate the forearm towards the ceiling. 12-15 repetitions, 3 sets per side. It is more demanding than the band because the load peaks right at the end of the range.
Prone Y-T-I. Lying face down on a bench or the edge of the bed, arm hanging. Lift forming the Y (arm at 45° from the head), the T (perpendicular to the body) and the I (along the body). 10 repetitions per letter, 2-3 sets. It strengthens the middle and lower trapezius, rhomboids and posterior deltoid.
Scapular retraction. Bring the shoulder blades together as if you were holding a pencil between them, hold for 5 seconds, release. 15-20 repetitions, 3 sets.
Band row. Pull the band towards you by bending the elbows and squeezing the shoulder blades together at the end of the movement. 15 repetitions, 3 sets.
Face pull. Band anchored at face height, pull towards your face while separating your hands and bringing the elbows back. 15 repetitions, 3 sets. It trains the external rotators and posterior stabilisers together.
Wall push-up plus. A push-up against a wall; at the end, push a little further while moving the shoulder blades apart. 15 repetitions, 3 sets. It activates the serratus anterior.
Phase 4: return to activity
This only makes sense if the previous phases have been clean and pain-free, and if you have an activity to return to (overhead sport, work with arms raised, the gym).
Plank with opposite shoulder tap. In a plank position, alternately tap the opposite shoulder without rotating the pelvis. 10 taps per side. It integrates shoulder control under load.
Functional and sport-specific progressions. Throws with a medicine ball, controlled decelerations, technical movements at increasing intensity. These should be built around the real movement, not copied from a list.
Rhythmic perturbations. Arm raised while a helper applies small unpredictable pushes in various directions. To be clear: this is an exercise to do with a physiotherapist, not with a well-meaning friend: the dosage of the pushes is the whole content of the exercise.
How to adjust the load (the part that decides the result)
If studies cannot tell a rich programme from an essential one, the factor that matters is probably not which exercise you choose, but how you load it over time.
- The right weight is the one that gets you to the end of the set with technique intact. If you cannot complete 15 clean repetitions, it is too heavy. For the cuff you usually stay between 0.5 and 3 kg, or on low-resistance bands.
- Increase gradually: step by step, but without trusting a fixed percentage: the 10% per week rule has been tested and did not reduce injuries compared with a normal progression. The signal to follow is how you react in the 24 hours afterwards.
- The next-day rule: moderate discomfort during the exercise is acceptable if the following day the shoulder is as it was. If the pain increases and stays for 24-48 hours, you loaded too much.
- Slow, controlled movement, especially in the return phase. Jerky movements shift the work onto the wrong muscles.
- Do not ignore a sharp, localised pain: it is not "the tendon remodelling", it is the signal to stop and review the exercise.
One clarification worth making, because the opposite is often read online: exercises do not repair a torn tendon. They improve the strength, control and load tolerance of the shoulder, and in many cases this is enough to remove the pain and recover function. This is a different thing from anatomical healing of the tear, and promising it would be wrong.
Equipment
Very little is needed. Bands in two or three different resistances (the set costs about the same as one session and covers months of progression), light dumbbells from 0.5 to 3 kg, and a surface to lie face down on: the edge of the bed is perfectly fine for the prone exercises. Nothing else is needed.
How often, and for how long
During active pain: daily mobility and isometrics, light strengthening every other day.
During the strengthening phase: 3 sessions a week with at least one rest day between them. Training the cuff every day at high intensity does not speed anything up.
For prevention: 2-3 sessions a week of 15-20 minutes, which fit well as a warm-up before upper-body training or as accessory work at the end.
On timings, it is better to be realistic: in studies, improvement is assessed over windows of 3, 6 and 12 months. Anyone who expects a change in two weeks usually stops just when the work would start to pay off.
The mistakes we see most often
- Loads that are too high. The cuff is made of small muscles: excessive weight leads to compensating with the deltoid and trapezius, that is, training exactly what does not need training.
- Compensating without noticing. If you shrug, tilt the trunk or lift the elbow away from your side during rotations, the work has gone elsewhere. A mirror or a phone video fixes the problem in a single session.
- Neglecting the scapula. It is probably the most costly mistake, because it is invisible: the shoulder keeps hurting and it is not clear why.
- Training only external rotation. The subscapularis is part of the cuff and needs loading too.
- Progressing too fast after a good week. It is the most common way to end up back at square one.
- Persisting for months with a picture that is not the right one. If after 6-8 weeks of work done properly nothing has changed, the problem is not consistency: it is the diagnostic hypothesis.
Rotator cuff: exercises to avoid
This is the question that always comes up, and the honest answer is a little disappointing: there is no list of exercises that are forbidden for everyone. It is not the movement itself that causes damage, it is the wrong load at the wrong moment. The same exercise that reignites symptoms in the irritable phase becomes useful six weeks later.
That said, there are four recurring ways of getting it wrong, and they are worth more than any list of names:
- Working overhead while the shoulder is still reactive. Military press, upright rows, lateral raises beyond 90° with pain: they are not banned for ever, but while movement above the shoulder hurts they should be postponed, not "pushed through".
- Aggressive stretching of the painful shoulder. Pulling to "free up" an irritated shoulder usually irritates it more. Mobility returns when the pain subsides, not the other way round.
- Behind-the-neck exercises: lat pulldown and military press behind the neck, which take the shoulder into maximum external rotation and abduction. They are very popular as joint-friendly exercises, but poorly tolerated if the cuff is symptomatic, and in any case they can be replaced with front variations without losing anything.
- A load that never changes. Staying on the yellow band for life is not caution, it is standing still: the cuff needs a load that increases, otherwise it does not adapt.
The practical rule I use, and which you will find in more detail below: discomfort that stays below 4-5 out of 10 during the exercise and settles within 24 hours is acceptable. Pain that rises, or that is worse the next day, says that the load was too much, not that the movement is wrong for ever.
After surgery: why the programme is not yours to write
After a cuff repair the sutured tendon goes through healing phases with precise biological timings, and the protocol depends on the size of the tear, the quality of the tissue and the surgical technique. The timings are set by the surgeon who operated.
On the debate between early and delayed rehabilitation, a systematic review of 20 randomised trials covering 1,841 patients found very few differences in pain and function, a small advantage for early start on the recovery of range of motion, and, more relevant, no increase in the risk of re-tearing with early mobilisation (odds ratio 1.05; 95% CI 0.64-1.75). The authors do note, however, a huge variability between protocols and a high or unclear risk of bias in most studies (Mazuquin et al., 2021).
In other words: early mobilisation, when it is prescribed, does not seem to put the repair at risk. It does not mean that you can decide on your own when to remove the brace or when to start active strengthening. If you are interested in the full pathway, we have detailed it in the guide to rotator cuff rehabilitation after surgery.
When it makes sense to get assessed
Beyond the urgent situations listed at the start, an assessment concretely changes things if:
- the pain has lasted more than 4-6 weeks with no tendency to improve;
- you have a documented tear and do not know which of the four pictures it falls under;
- the exercises worsen your symptoms instead of settling them;
- you have been offered an operation and want to understand whether your case resembles those in which rehabilitation works;
- you have been operated on and the post-operative programme is not clear.
It serves two purposes: putting the right picture under the right name, on which you will find more detail in the guide on symptoms, diagnosis and treatment of a rotator cuff tear, and calibrating the load to you, which is the variable that ultimately decides the result.
You might also be interested in:
Band Pull-Apart: The Essential Exercise for Healthy Shoulders
Trapezius Knots: Causes, Symptoms and Exercises to Release the Tension
Core Strengthening Exercises: A Complete Guide to Abs, Back and Stability
Stay in touch with us
Join our dedicated Facebook group: we answer all patients, share practical advice and clear up the most common questions.
Wondering how much it would cost to be followed? The spend depends on how many sessions are needed, not on the single rate: you will find the figures in How much physiotherapy for the shoulder costs.
Need a consultation?
Let's see whether a course makes sense for you: a few questions about your case, you register, and you see who will follow you: a physiotherapist registered with the Albo, whom you can write to in chat whenever you like. If you would rather talk it through, the first consultation is free, lasts twenty minutes and commits you to nothing: they will arrive having already read what you wrote, and together you decide whether and how to go on.
If the pain is more at the front than at the side, it may not be the cuff: see Biceps tendon: pain at the front of the shoulder.
Sources
- Page MJ, Green S, McBain B, et al., Manual therapy and exercise for rotator cuff disease, Cochrane Database of Systematic Reviews, 2016, pubmed.ncbi.nlm.nih.gov/27283590
- Hopewell S, Keene DJ, Marian IR, et al., Progressive exercise compared with best practice advice, with or without corticosteroid injection, for the treatment of patients with rotator cuff disorders (GRASP), The Lancet, 2021, pubmed.ncbi.nlm.nih.gov/34265255
- Littlewood C, Bateman M, Brown K, et al., A self-managed single exercise programme versus usual physiotherapy treatment for rotator cuff tendinopathy: a randomised controlled trial (SELF), Clinical Rehabilitation, 2016, pubmed.ncbi.nlm.nih.gov/26160149
- Karjalainen TV, Jain NB, Heikkinen J, et al., Surgery for rotator cuff tears, Cochrane Database of Systematic Reviews, 2019, pubmed.ncbi.nlm.nih.gov/31813166
- Kukkonen J, Ryösä A, Joukainen A, et al., Operative versus conservative treatment of small, nontraumatic supraspinatus tears in patients older than 55 years: over 5-year follow-up of a randomized controlled trial, Journal of Shoulder and Elbow Surgery, 2021, pubmed.ncbi.nlm.nih.gov/33774172
- Moosmayer S, Lund G, Seljom US, et al., At a 10-year follow-up, tendon repair is superior to physiotherapy in the treatment of small and medium-sized rotator cuff tears, The Journal of Bone and Joint Surgery (Am), 2019, pubmed.ncbi.nlm.nih.gov/31220021
- Gutman MJ, Joyce CD, Patel MS, et al., Early repair of traumatic rotator cuff tears improves functional outcomes, Journal of Shoulder and Elbow Surgery, 2021, pubmed.ncbi.nlm.nih.gov/33774173
- Mazuquin B, Moffatt M, Gill P, et al., Effectiveness of early versus delayed rehabilitation following rotator cuff repair: systematic review and meta-analyses, PLOS ONE, 2021, pubmed.ncbi.nlm.nih.gov/34048450
- Minagawa H, Yamamoto N, Abe H, et al., Prevalence of symptomatic and asymptomatic rotator cuff tears in the general population: from mass-screening in one village, Journal of Orthopaedics, 2013, pubmed.ncbi.nlm.nih.gov/24403741
- Hinsley H, Ganderton C, Arden NK, Carr AJ, Prevalence of rotator cuff tendon tears and symptoms in a Chingford general population cohort, and the resultant impact on UK health services, BMJ Open, 2022, pubmed.ncbi.nlm.nih.gov/36100305
Last source check: August 2026.
