Rounded shoulders come with the tidiest explanation in the whole posture story: the chest muscles get short from hunching, the upper back muscles get weak from not being used, the shoulders roll forward, and pain follows. Stretch the front, strengthen the back, and you are fixed.
It is a good story. The measurement it depends on does not distinguish people with shoulder symptoms from people without, and the position it blames does not track the condition it is blamed for.
The short answer
- The standard test for a short pectoralis minor has poor diagnostic accuracy for shoulder symptoms.
- Scapular orientation and subacromial shoulder pain do not show a consistent relationship in the reviews.
- Non-radiographic measures of upper-back curvature have limited reliability, so the reading itself is soft.
- Chest and upper-back work is still worth doing — for range and capacity, not as correction.
- What has randomised trials behind it in office workers is strength, in very small doses.
Where the rounded shoulders story starts
The mechanism is anatomically reasonable. The pectoralis minor attaches from the ribs to the coracoid process of the shoulder blade, so a shorter one would tilt the blade forward and down. Borstad and Ludewig demonstrated in the Journal of Orthopaedic & Sports Physical Therapy in 2005 that people with a shorter resting pectoralis minor length showed different scapular kinematics during arm elevation than people with a longer one.
So the first link — short pec minor, different blade movement — has real measurement behind it. Everything after that is where it comes apart.
Does the test find anything
Lewis and Valentine examined the pectoralis minor length test in BMC Musculoskeletal Disorders in 2007, in people with and without shoulder symptoms. They found the test could be repeated reasonably by the same examiner and had poor diagnostic accuracy: it did not usefully separate the symptomatic group from the asymptomatic one.
Ratcliffe and colleagues went at the wider claim in the British Journal of Sports Medicine in 2014, systematically reviewing whether there is a relationship between subacromial shoulder pain and scapular orientation. They found no consistent relationship. Studies disagreed on which way the blade sat, and on whether it sat differently at all.
Barrett, McCreesh and Lewis reviewed the non-radiographic ways of measuring thoracic curvature in Manual Therapy in 2014 and found reliability and validity varying widely between methods. So even the measurement of "how rounded" a person is depends heavily on who is measuring and how.
Grade: weak. The mechanism is plausible, the test does not identify symptomatic people, the position does not track the condition, and the measurement is soft. That is four links in a chain and none of them holds a load.
What about pain
The general posture-and-pain evidence applies here as it does everywhere else in this cluster.
Richards and colleagues found no association between measured sitting neck posture clusters and neck pain in more than 1,000 adolescents, in Physical Therapy in 2016. Swain and colleagues, reviewing reviews in the Journal of Biomechanics in 2020, found no consensus that spinal posture causes low back pain. Christensen and Hartvigsen, in the Journal of Manipulative and Physiological Therapeutics in 2008, found no consistent association between sagittal spinal curvature and health.
The counterweight, stated at full strength: Mahmoud and colleagues' 2019 meta-analysis in Current Reviews in Musculoskeletal Medicine did find a small-to-moderate association between forward head posture and neck pain in adults. It is cross-sectional and cannot establish direction, and shoulder position was not what it measured. Forward head posture works through it.
What is worth doing anyway
Drop the correction frame and most of the same exercises survive, with better reasons attached.
Chest and front-of-shoulder range. A doorway pec stretch — forearm on the frame, elbow around shoulder height, step through, 30 seconds each side — takes the shoulder into external rotation and extension, which a screen day removes. Do it because the range is worth having, not because your chest is pulling you over.
Thoracic extension and rotation. The upper back is the segment a desk flattens most completely. A seated tall rotation, 40 seconds each side, and an extension over the back of a chair are the two cheapest ways to visit it — both are in desk stretches with their holds. Limber's four-week desk track is built around exactly this: twelve sessions across the upper back, neck and shoulders, from gentle to full extension.
Strength for the shoulder blades and the neck. This is the part with evidence. Andersen and colleagues, in Pain in 2011, gave office workers with frequent neck and shoulder pain two minutes a day of progressive resistance training and measured a real reduction over ten weeks. Their 2008 trial in Arthritis & Rheumatism found specific strength training for the painful muscles outperformed general fitness training. Blangsted and colleagues' one-year randomised trial in the Scandinavian Journal of Work, Environment & Health in 2008 found specific resistance training reduced symptoms where an all-round activity programme did not.
Grade: moderate, and the strongest thing in this area. Randomised, workplace, real clinical outcomes. It is also not stretching, which is worth admitting on a page written by a mobility company.
And move often. Two minutes an hour. The break evidence is weak but consistent and the cost is nothing.
What to stop doing
Stop assessing yourself in a mirror. Self-judgement of habitual posture is unreliable and the act of checking changes the thing being checked.
Stop wearing a posture brace expecting it to change anything. There is no serious trial evidence that they reduce symptoms, and their premise is the link that keeps failing. They also do work the muscles would otherwise do, which cuts against the one finding here with decent support.
Stop reading rounded shoulders as damage. Shoulder position varies enormously in people who feel fine, and the reviews cannot even agree that it differs between people who hurt and people who do not.
If your shoulder hurts rather than merely sitting forward, that is a different page. Pain that wakes you at night, arm weakness, numbness or pins and needles down the arm, pain following a fall, or a shoulder that has lost range suddenly needs a clinician who can examine you. Nothing here diagnoses anything and nothing here is treatment.
Measure the range, not the shape
The useful reading here is not how rounded you look. It is how much reach you have, and it is easy to take.
The shoulder mobility test is the Apley scratch, scored in five bands, left and right kept apart: one arm over the top and one under, reaching behind the back, measured in hand-widths. More than two hand-widths apart reads 10, one to two reads 30, under one reads 55, fingertips touching reads 75, fingers overlapping reads 95.
Two honest notes about it. It measures three joints at once and cannot say which one is short — the top arm is doing flexion and external rotation, the bottom arm extension and internal rotation, and the blade has to slide on the ribs for either to travel. A stiff upper back closes the gap between the hands as surely as a stiff shoulder does. And the smallest gap it can show between your sides is 20 to 25 points, while Limber calls 12 points an imbalance — so any one-band difference between sides is flagged, and on a banded test that means re-measure, not finding.
The Range Score composes that reading with the rest of the battery, at 60% of the shoulders area — the wall angel test that would supply the other 40% is not built, so the page prints the shortfall rather than quietly renormalising. The cluster is desk and posture.
Questions
How do I fix rounded shoulders?
You can improve chest and shoulder range and build strength through the upper back, and both are worth doing. Whether that changes your resting shoulder position durably is largely unstudied, and whether the position matters is the part the evidence does not support — the review of scapular orientation and shoulder pain found no consistent relationship.
Are rounded shoulders caused by tight pecs?
That is the standard explanation and it is weaker than it sounds. A shorter pectoralis minor does correlate with different scapular movement, but the clinical test for a short pec minor has poor accuracy at separating people with shoulder symptoms from people without. Plenty of people with forward-sitting shoulders have no symptoms at all.
Do rounded shoulders cause shoulder pain?
The systematic review that asked directly whether subacromial shoulder pain relates to scapular orientation found no consistent relationship. Shoulder position varies widely in people who feel fine, so it is not a reliable explanation for a painful shoulder.
How long does it take to change shoulder posture?
Studies that have shifted posture measurements ran eight to sixteen weeks, mostly on head position rather than shoulders, and rarely followed people afterwards. Given how weak the link to symptoms is, a better target is range and strength — both of which respond over three to six weeks and can be measured.
Are posture correctors worth wearing?
No serious trial evidence supports them for symptoms, and their premise rests on the posture–pain link that keeps failing its tests. They also substitute for muscular work, which runs against the one well-supported finding here: small doses of resistance training reduce neck and shoulder pain in office workers.
What is the single best exercise for this?
If you want range, a doorway pec stretch, 30 seconds a side. If you want the thing with randomised trials behind it, a light lateral raise — two minutes a day was the dose that reduced neck and shoulder pain in office workers over ten weeks. The second has better evidence than the first.