The 90/90 stretch is the one seated position that loads both hip rotations at the same time. The front leg turns outward, the back leg turns inward, and switching between sides gives each hip a turn at each direction. Nothing else in a normal routine does that in one shape.
It is also the position most often set up wrongly, and set up wrongly it stretches the lower back while the hips do nothing. Three errors account for almost all of it, and they are all about where your pelvis is.
The short answer
- Front leg bent at right angles in front, back leg bent at right angles out to the side. Sit tall.
- The front hip is in external rotation, the back hip is in internal rotation. One position, both directions.
- If you cannot sit upright in it, sit on something. Rounding backwards hands the work to the lumbar spine.
- In the app it is M012 90/90 Hip Switch — dynamic, 45 seconds, level 2, on a mat, cleared for warm-ups.
- No trial of this position exists. Grade: absent. The transferable evidence is about dose, not about the shape.
- Nothing in Limber's battery reads rotation, so no score here will show it working.
Setting up the 90/90 stretch
Sit on the floor, on a mat, with room around you.
Front leg. Bring one leg in front of you, bent so the hip and the knee are each at about a right angle, with the shin roughly parallel to your chest. That hip is now flexed and turned outward.
Back leg. Take the other leg out to the side, also bent at about a right angle at the hip and the knee, with the shin behind you. That hip is now turned inward. This is the half nobody trains anywhere else.
Sit tall. The spine long, the chest lifted, both sitting bones on the floor. If your lower back is rounding backwards, stop and read the next section before you do anything else.
Switch. Lift both knees, rotate them across to the other side, and settle into the mirror image. Slowly, from the hips, with the chest staying up. One switch per breath cycle, six to twelve of them, about 45 seconds in total.
The sensation belongs deep in the outside of the front hip and across the front and inside of the back one. A pinch deep in the hip socket is a signal to ease off, which is the app's own instruction for this move.
Both rotations in one position
Here is why the shape is worth the trouble.
Almost every hip stretch people do — the pigeon, the figure-four, the butterfly — loads external rotation. Internal rotation is trained by nearly nothing, and it is the direction a hip needs to stand on one leg, change direction, and sit into a squat without the pelvis rolling under. Hip internal rotation is the article on that direction and on how to check it, since no test here reads it.
In the 90/90 you get both at once, on opposite sides, and the switch alternates them. That is efficient in a way most hip work is not: two directions, two hips, one position, 45 seconds. Compare that with the rest of the hip catalogue, where each direction usually needs its own move and its own 40 seconds.
The three setup errors
Each of these turns the position into something else.
Sitting behind your hips. This is the big one. If your hamstrings and back hip will not let you sit upright, the pelvis rolls backwards, the lower back rounds, and the tension you feel is lumbar rather than hip. You have found a lower-back stretch with your legs in an unusual shape. The fix costs nothing: sit up on a folded blanket, a cushion or a couple of books so the hips are higher than the knees, or put your hands on the floor behind you and lift the chest. Height first, depth later.
The back knee drifting in. The back leg wants to slide toward your body, because that closes the angle at the back hip and removes the internal rotation demand — which is the thing you came for. Keep the back thigh out to the side and the knee at roughly a right angle. If it will not stay there, you are asking for more inward range than that hip currently has; take less.
The trailing hip lifting. Watch the back hip. When it peels off the floor, the rotation has stopped coming from the joint and started coming from the whole pelvis tipping. This is the same substitution that ruins a hip flexor stretch, one joint along. Keep both sitting bones down, accept a smaller shape, and the hip gets the work instead of the pelvis.
A rule that covers all three: the position you can sit upright in is worth more than the position you can reach. Depth in this shape is easy to fake and the fake version trains nothing you wanted.
Why the app makes it a switch
In Limber the move is M012 "90/90 Hip Switch", and its record is specific: dynamic, level 2, a 45-second default, on a mat, cleared for warm-ups, ease off at any pinch deep in the hip socket. It is not filed as a hold and it is not per side, because the switch alternates the sides for you.
That choice suits the position. Most people cannot yet sit upright in the shape, and a long static hold in a shape you cannot sit upright in is 45 seconds of lumbar flexion under a load you did not intend. Moving through it changes the terms. You pass through the end of each hip's range instead of camping at it, both hips work in the same 45 seconds, and the intensity is set by how far you rotate rather than by how long you can tolerate a position you are already losing.
It also makes the move legal in a warm-up, where a hold at end range would not be. The session grammar is one or two dynamic moves at the front, a main block, and one long calm hold to finish. The switch belongs at the front.
The evidence, and its absence
Say the grade out loud, because it is unusual on this site to have so little to grade.
No trial of the 90/90 position exists. Grade: absent. Nobody has randomised people to it, nobody has compared it with another rotation position, and nobody has measured what it does to hip rotation range over weeks. The case for it is anatomical — the shape plainly puts one hip in each rotation — and anatomical plausibility is where a great many exercises that turned out to do nothing also started.
What can be borrowed is general. Bandy and Irion, in Physical Therapy in 1994, found 30-second holds beat 15 and matched 60 over six weeks, and their 1997 follow-up in the same journal found one hold a day matched three. Thomas and colleagues, in the International Journal of Sports Medicine in 2018, pooled the dose literature and concluded that total weekly time is the variable that matters, not the length of a single hold. Grade: solid, and narrow — hamstrings, young healthy adults, static holds. Applying it to a dynamic hip switch is an extrapolation and this page will not pretend otherwise.
And the reading you will not get: nothing in Limber's battery measures rotation at the hip. The five tests are the toe touch, knee-to-wall, shoulder reach, hip flexor test and deep squat. So there is no score on this site that will rise because you did this position for six weeks, and any page that implies one is wrong.
When the near hip stops hard
Some people meet a hard stop in the front hip within a few centimetres of settling in, and no amount of work moves it.
Frank and colleagues reviewed imaging of people without symptoms in Arthroscopy in 2015 and found cam-type morphology — extra bone at the head-neck junction of the femur, which mechanically blocks deep flexion and inward rotation — in roughly a third of asymptomatic volunteers and over half of asymptomatic athletes. Grade: reasonably solid for the prevalence, unsettled for what it means.
The practical version is a sensation you can learn. A stretch is felt as length along a line and usually softens if you stay there. Bone contact is a hard pinch at the front of the hip or the groin, it turns up at the same place every time, and staying there does nothing to it except irritate the joint. Change the angle rather than the effort: less depth, more height under the hips, a slightly different foot position.
Pain that persists, wakes you at night, radiates into a limb, comes with numbness or weakness, or followed a fall or an injury is a matter for a clinician who can examine you. Nothing here diagnoses anything and nothing here is treatment.
Where to start
Sit up on a cushion, whatever your range. Do six to twelve switches, roughly 45 seconds, four or five days a week, at the front of a session while you are warm. Judge it on whether you need less height under your hips in a month, not on a score.
For the readings that do exist, the hip flexor test reads the front of each hip in three bands and the deep squat test reads the whole lower body in four. Neither is a rotation measure and both are still worth having, because they are the two nearest a hip and they are free. The hip mobility test runs the pair in one sitting and adds a seated rotation check it compares side against side without scoring, which is the nearest thing to a reading this position will get. Take them, write down the date, take them again in three weeks under the same conditions.
The back of the hip is covered in glute stretches, the six directions and their moves in hip mobility exercises, and the whole cluster sits under hips and legs.
Questions
What is the 90/90 position?
Sitting on the floor with one leg bent at right angles in front of you and the other bent at right angles out to the side. The front hip is turned outward and the back hip is turned inward, so a single position loads both rotations at the same time. Switching the legs across to the mirror image gives each hip a turn at each direction.
Why does my lower back hurt in the 90/90?
Almost always because you are sitting behind your hips. If the pelvis rolls backwards, the lumbar spine rounds and takes the tension that the hips were supposed to take. Sit up on a folded blanket, a cushion or a couple of books so that the hips are higher than the knees, and the same shape starts working on the joints you meant.
Should I hold it or switch sides?
The app treats it as a dynamic move: six to twelve switches over about 45 seconds, one per breath cycle. That suits a shape most people cannot yet sit upright in, because it avoids a long hold at end range while still moving both hips through both rotations. If you can sit tall in the position comfortably, adding a still hold afterwards is a reasonable extension.
Does this position improve hip internal rotation?
No trial has tested it, so the honest grade is absent. The shape unmistakably places one hip in internal rotation, and the general dose research suggests total weekly time matters more than the length of any single session, but neither of those is a study of this position. Treat it as well-reasoned rather than demonstrated.
Can I do the 90/90 in a warm-up?
Yes. It is one of the hip moves cleared for warm-ups in the app, because it is a switch rather than a long hold at the end of a range. That is the general rule the session grammar follows: dynamic moves at the front, longer static holds in the main block and at the finish.
Why does the front of my hip pinch when I sit into it?
A pinch deep in the socket is usually contact rather than length, and the app's own instruction for this move is to ease off when it appears. Extra bone at the head-neck junction of the femur is common in people with no symptoms at all, and it blocks the inward direction mechanically. Take less depth, raise your hips, and if it catches or lingers afterwards see a clinician who can examine the joint.