Squat Thoracic Rotation
Sit in a deep squat with your torso as upright as it will go and both arms between your knees. From there, open one arm upwards and follow your hand with your eyes while the lower hand pushes the opposite knee outwards. Then back to the middle and over to the other side. If you can't get into a deep squat yet, raise your heels or support yourself with your lower hand on the floor.
The squat thoracic rotation turns the ribcage while the hips sit in deep flexion. The deep squat is the whole point: standing up, you always turn a fair bit through the hips as well, and down in the squat there is barely room for that. The rotation ends up where it is meant to be.
Your lower hand on the opposite knee gives you the anchor to turn against, and it takes the thigh out a little at the same time.
You need nothing but floor space. If the deep squat is not there yet, put your heels on a low block or take support with your lower hand on the floor — the rotation is the same.
From practice: I put this in the warm-up before anything overhead. If you have come straight from the desk or the car, keep the first two reps deliberately small and slow. That is what I see in the studio.
Starting position
Drop into a deep squat, as low as you can go without pain. Your feet are about shoulder-width apart, your heels down if possible, and your torso is as upright as it will go in this position.
Your elbows are now between your knees, both arms hanging down inside them. From there, open one arm upwards and follow your hand with your eyes. Your lower hand stays on the floor, or rests against the inside of the opposite knee and pushes it outwards. Then come back to the middle and open to the other side.
At first only rotate as far as you can go without feeling a pull, and move into it slowly. If you go into the end position too fast while you are still cold, the ribcage can lock up on you.
If the deep squat itself is not available yet: put your heels on a low block of three to five centimetres, or sit down so that your lower hand supports you on the floor.
The three cues
- Elbows between your knees, torso as upright as it will go
- Open one arm upwards and follow your hand with your eyes
- The lower hand pushes the opposite knee outwards at the same time
- Rotate slowly, only as far as you can go without a pull
Common faults
| Fault | How to fix it |
|---|---|
| The rotation comes from the hip instead of the ribcage | Your pelvis stays where it is in the squat and your feet stay flat. Turn deliberately from higher up and leave your knees where they are. |
| The torso collapses forward | Draw yourself tall for a moment before every rep. Very little rotation comes out of a rounded back. |
| Going into the end position too fast | Move in slowly, especially while you are cold. Fast and far at once is how the ribcage shuts down on you. |
| The lower hand does not push | Without pressure on the opposite knee there is no anchor and you get less range. Put the hand on the knee and push outwards. |
| The heels lift and your weight tips forward | Put your heels on a low block, or take support with your lower hand on the floor. Both are a full version of the exercise. |
Variations
| Variation | When to use it |
|---|---|
| Heels raised or supported on the floor | Heels on a low block, or your lower hand supporting on the floor. The squat does not have to be deep. |
| Free deep squat | Heels down, elbows between the knees, alternating sides. |
| Hold the end position | Hold the open position for two or three breaths. |
Reps and load
Five to eight rotations per side, alternating. Slowly in and slowly back.
It fits into the warm-up before upper-body pushing and pulling, and before anything overhead. Alongside the bird dog and the shoulder dislocate it makes a short warm-up for the shoulders and trunk.
Which muscles do the work
| Muscle | Role | What it does here |
|---|---|---|
| External oblique M. obliquus externus abdominis | Prime mover | Turns the ribcage towards the opposite side |
| Internal oblique M. obliquus internus abdominis | Prime mover | Turns the ribcage towards its own side, paired diagonally with the opposite external oblique |
| Latissimus dorsi M. latissimus dorsi | Prime mover | Carries the rotation through the arm and shoulder girdle |
| Deep rotators Mm. rotatores, Mm. multifidi | Synergist | Rotate the thoracic vertebrae relative to one another, segment by segment |
| Erector spinae M. erector spinae | Stabiliser | Hold the torso upright in the squat |
| Gluteus maximus M. gluteus maximus | Stabiliser | Turns the thigh outwards as the hand pushes the knee |
| Hip flexors M. iliopsoas, M. rectus femoris | Stabiliser | Hold the hip in deep flexion so the pelvis stays put |
The rotation runs through a diagonal sling: the obliques on both sides work against each other, and the latissimus carries it through the arm and shoulder girdle. Spinal erectors, glutes and hip flexors hold the squat you turn out of.
Explained in full
Andreas explains why rotation gets lost over a day spent sitting, then shows the deep squat with elbows between the knees, the opening rotation with the eyes following the hand, and the entry versions with the heels raised and with the lower hand supporting on the floor. · 2:11
Seeing it done right is not the same as doing it right
Video, cues and variations take you a long way. What they cannot give you is an outside eye at the moment the movement isn't quite there yet. That is what training with me is for — in the studio or in a workshop.
See the workshopsNew exercise pages are added all the time — the newsletter keeps you posted.
Sources
Bayram K, Kaya DÖ. (2026). Effects of adding thoracic spine exercises to routine soccer training on spinal alignment and mobility in professional male soccer players: a randomized controlled study. BMC Sports Sci Med Rehabil 18(1):207. DOI: 10.1186/s13102-026-01633-9 — A randomised study in 42 professional male footballers (20.9 ± 3.4 years), allocated 1:1 to six weeks of thoracic mobility exercises three times weekly on top of training, or training alone. Bilateral thoracic rotation angles improved markedly in the exercise group (Cohen's d = 1.00 to 1.46), while the actual primary outcome — thoracic kyphosis in the sagittal view — did not change. The basis for saying targeted rotation work raises measurable rotation within six weeks, and for making no claims about posture here. The sample is young competitive athletes, the control group received no sham intervention, rotation was measured with a smartphone app, and the trial was registered retrospectively. (Evidence type: RCT)
Webber SC, Porter MM. (2022). Effect of a supervised stretching program on neck, shoulder, and trunk range of motion in older women. Can J Aging 41(3):297–303. DOI: 10.1017/S0714980821000350 — 48 community-dwelling women (75 ± 3 years) were allocated to supervised upper-body stretching (n = 15) or lower-body power training as control (n = 33), 45 minutes twice weekly for twelve weeks. Only shoulder range of motion improved beyond the controls (a 33 per cent gain); neck and trunk range did not change. Included as the counterweight to the assumption that mobility work reliably increases trunk rotation: twelve weeks of supervised stretching was not enough in this group. A considerably older sample than the target group, women only, badly unbalanced with 15 people in the exercise arm, and static stretching rather than active rotation drills. (Evidence type: RCT)
Fan JZ, Liu X, Ni GX. (2014). Angular velocity affects trunk muscle strength and EMG activation during isokinetic axial rotation. Biomed Res Int 2014:623191. DOI: 10.1155/2014/623191 — Surface EMG of external and internal oblique plus latissimus dorsi bilaterally in 24 healthy young men during isokinetic trunk rotation to right and left at 30, 60 and 120 degrees per second. The working muscles were the contralateral external oblique together with the ipsilateral internal oblique and latissimus; as speed rose, torque and agonist activity fell while the antagonists braked more. The basis for the diagonal sling in the muscle table and for the slow execution. Young men only, measured on a dynamometer standing against resistance rather than in a deep squat, and the deep rotators cannot in principle be reached with surface measurement. (Evidence type: EMG)
Doriot N, Wang X. (2006). Effects of age and gender on maximum voluntary range of motion of the upper body joints. Ergonomics 49(3):269–281. DOI: 10.1080/00140130500489873 — A comparison of maximum voluntary range of motion across 13 seated upper-body movements in 41 people: 22 aged 25 to 35 against 19 aged 65 to 80. The effect of age differed by joint, and the largest losses were at the neck and trunk, explicitly for trunk lateral flexion and trunk rotation; elbow and wrist showed no age difference. The basis for saying trunk rotation is among the movements that decline most. The abstract gives the direction but no degree values, the sample is small, the comparison skips the 40-to-60 group, and trunk rotation was measured as a whole rather than isolated to the thoracic spine. (Evidence type: Querschnitt)
Heneghan NR, Baker G, Thomas K, Falla D, Rushton A. (2018). What is the effect of prolonged sitting and physical activity on thoracic spine mobility? An observational study of young adults in a UK university setting. BMJ Open 8(5):e019371. DOI: 10.1136/bmjopen-2017-019371 — An observational study in 92 people (18 to 30 years), grouped by self-report into sitters (over 7 hours a day, under 150 minutes of activity a week), physically active and low activity. Thoracic mobility in the heel-sit position was 64.8 degrees in sitters, 75.0 degrees in the active group and 68.4 degrees in the low-activity group, with clear differences between sitters and both other groups. The basis for saying a seated day goes with less thoracic rotation. Cross-sectional with no cause and effect, 18- to 30-year-olds rather than the target group, activity and sitting self-reported only, and the heel-sit position is not a deep squat. (Evidence type: Querschnitt)
Hemmerich A, Brown H, Smith S, Marthandam SSK, Wyss UP. (2006). Hip, knee, and ankle kinematics of high range of motion activities of daily living. J Orthop Res 24(4):770–781. DOI: 10.1002/jor.20114 — Three-dimensional hip, knee and ankle kinematics captured with electromagnetic tracking in 30 healthy people with a mean age of 48.2 ± 7.6 years, six trials each of squatting, kneeling and sitting cross-legged. Squatting with raised heels required a mean of 157 ± 6 degrees of knee flexion, squatting with flat heels up to 95 ± 27 degrees of hip flexion — the wide spread shows how much the individual demand varies. The basis for offering the entry versions at all. The closest age match among these sources, but the participants come from a life of habitual squatting and are therefore more mobile than the Central European average; no ankle threshold is reported and there are no thoracic data. (Evidence type: Biomechanics)
Ghasemi M, Emami M, Mohammadi Yaghoubi U. (2026). Heel elevation increases ankle and knee range of motion during squatting in healthy adults: a systematic review with meta-analysis. Sports Biomech 25(7):1022–1038. DOI: 10.1080/14763141.2026.2619893 — A review with meta-analysis of 14 studies and 177 healthy participants on the effect of heel elevation during squatting. Elevation increased range of motion at the ankle (mean difference 4.33 degrees) and knee (4.94 degrees), but not at the hip or trunk; at the ankle, gains appeared only above 2.5 centimetres, and meta-regression suggested that a higher heel tends to reduce trunk range. The basis for framing raised heels as a way into the deep squat rather than a route to more rotation. A small pooled sample, age and training status not broken out in the abstract, standing squat movements rather than a held deep squat, and thoracic rotation does not feature. (Evidence type: Meta-analysis)
Johnson KD, Kim KM, Yu BK, Saliba SA, Grindstaff TL. (2012). Reliability of thoracic spine rotation range-of-motion measurements in healthy adults. J Athl Train 47(1):52–60. DOI: 10.4085/1062-6050-47.1.52 — In 46 healthy volunteers (23.6 ± 4.3 years), five clinical techniques for thoracic rotation were tested across two days, including seated rotation and half-kneeling. All returned reliability values of 0.84 to 0.95, with a measurement error of 0.8 to 2.3 degrees and a smallest detectable change of 2.1 to 6.3 degrees. This supports only the claim that thoracic rotation can be measured reproducibly and that changes below roughly six degrees may be measurement noise in an individual — not normative values, which the study does not report. A young sample without the target group, and none of the tested positions is a deep squat. (Evidence type: Messmethodik)
A note on context: EMG studies show which muscles an exercise recruits — that is a good reason to pick it, but it does not replace adjusting things to the person in front of you. See a doctor for persistent or severe pain.