Almost all hip flexor stretches are performed in a way that stretches the lower back instead. The position looks correct from the outside, the sensation arrives in roughly the right place, and the hip barely moves. The difference is the pelvis, and it is the only technical point on this page that matters.
The moves themselves are simple and there are only three of them in this catalogue. What separates a hip flexor stretch that works from one that does not is whether the pelvis stays put while the thigh travels behind you.
The short answer
- The hip flexors cross the front of the hip; to lengthen them the thigh must travel behind the body.
- If the pelvis is free to tip forward, the lower back extends and the hip does not.
- Tuck the tailbone under, squeeze the back glute, ribs down, then move forward a few centimetres. Not more.
- Only three of the catalogue's 56 moves work the hip flexors: M003, M015 and M020.
- The test this site uses to read them has a published validity problem, printed below in full.
Hip flexor stretches that work
Set the pelvis first, then take the range. In that order, every time.
Get into half kneeling. Back knee on a mat, front foot flat, both hips facing forward. Most people stop here and lunge, which is the error.
Tuck the tailbone under. Think of the front of the pelvis lifting toward the ribs. The lower back flattens slightly. You will feel the front of the back hip take tension before you have moved at all — that is the stretch starting.
Squeeze the glute on the back leg. This holds the tuck and adds a small amount of hip extension from the muscle that produces it.
Keep the ribs down. Flaring the ribs is the same error as tipping the pelvis, one segment higher.
Now move forward, a little. Two or three centimetres of travel is usually enough once the pelvis is set. If you can slide a long way forward comfortably, the pelvis has probably let go.
Hold for 40 seconds. Breathe in for four and out for six. The sensation belongs across the front of the hip and the top of the thigh. A pinch deep in the front of the hip socket is a signal to ease out, not to press on.
What the pelvis has to do
Here is the mechanism in one paragraph, because it explains every version of the error.
The hip flexors run from the lumbar spine and the pelvis to the top of the thigh bone. Lengthening them needs the two ends to separate, which means the thigh going back relative to the pelvis. If the pelvis tips forward as the thigh goes back, the two ends keep their distance and nothing has been stretched. The travel has to come out of the joint. The lower back is the nearest willing substitute, and it is the reason so many people finish a lunge stretch with an aching back and an unchanged hip.
The same substitution corrupts the measurement. The test used to read hip flexor length has exactly this problem, and it is our test.
The three moves in the app
The catalogue holds 56 moves across ten body areas, and three of them are filed to the hip flexors. That is a small number and it is the true one.
| Move | Type and hold | Notes |
|---|---|---|
| M003 Half-Kneeling Hip Flexor Stretch | Hold, 40 s per side | Mat; contract-relax capable |
| M015 Standing Quad Stretch | Hold, 30 s per side | Wall; reaches the quadriceps too |
| M020 Lizard Lunge | Hold, 40 s per side | Mat; filed to hips and hip flexors |
M003 is the reference position and the one the pelvis instructions above describe. M015 is the standing version, one hand on a wall, heel drawn toward the backside — the same tailbone tuck applies, and without it the move becomes a knee stretch. M020 Lizard Lunge is the deepest of the three: front foot outside the hands, forearms toward the floor if they reach. It is filed to the hips as well as the hip flexors because it loads both.
There is no couch stretch in this catalogue. M003 and M015 are the nearest positions to it, and the couch stretch covers what the elevated-foot version adds and what it costs.
Adding a contract-relax round
Of the three, only M003 is flagged for contract-relax work. The grammar the app uses is fixed and worth following exactly.
Take the position and hold it normally. Then press the back knee gently down into the floor — contracting the hip flexor you are stretching — at about 40% of your effort, for 6 seconds. Release for 2 seconds. Then deepen on the exhale and hold the new position for 15 to 20 seconds. Two or three rounds, no more.
Two rules go with it. Never in a warm-up: this belongs in the main block of a session, on a hip that is already warm. And 40% means 40%. A maximal contraction near end range is a good way to make a hip sore and a poor way to make it longer.
The instrument has a known flaw
This is the finding that goes against us, and it goes first.
Vigotsky and colleagues, in PeerJ in 2016, examined the modified Thomas test and concluded that it is not a valid measure of hip extension unless pelvic tilt is controlled. The reason is the mechanism described above: if the pelvis tips, the thigh drops, and the thigh dropping is what the test scores. A tilted pelvis produces a better reading from a hip that has not changed.
Grade: solid, and it is our test. The hip flexor test on this site is a Thomas-style reading, taken lying on your own back, judging your own hanging thigh, with nobody holding your pelvis. Everything Vigotsky and colleagues object to applies to it.
So why is it here? Because the alternative is no reading at all, and a repeatable reading you take against yourself under the same conditions still carries information even when its absolute validity is poor. The protocol asks you to pull the opposite knee in hard and keep the lower back flat precisely to limit the tilt, and the page says what the tilt does. What it cannot claim is that the band you scored is a measure of hip extension in degrees. It is not.
Three bands, thirty points
The second honest limit is arithmetic, and it is worse than the first.
The hip flexor test scores in three bands: 25, 60 and 90. The thigh floating above the bench line reads 25. Level with the bench reads 60. Hanging below it reads 90. Those are the only three values the instrument can produce.
The steps between them are 35 points and 30 points. So nothing smaller than a 30-point change can be recorded at all. Six weeks of good work that genuinely improved your hip by a few degrees will show as no change whatsoever, because the instrument has nowhere to put it.
The same arithmetic bites on the left-right comparison. Limber calls a difference of 12 points or more between your sides an imbalance, and the engine adds one extra move on the low side each session until the gap falls under 6. But the smallest gap this test can express is 30 points. Every possible one-band difference already exceeds our own threshold. On this test, therefore, a one-band difference means re-measure, not finding — take it again on another day, with the same setup, before you believe it. The Range Score restates the gap rather than averaging it away, which is the right behaviour and does not rescue the underlying coarseness.
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
Take the hip flexor test today, both sides, and write down the date and the two bands — or take it inside the hip mobility test, which puts it next to a 30-second squat hold and a seated rotation check compared side against side rather than scored. Then do M003 for 40 seconds a side, four or five days a week, with the tailbone tucked. Add contract-relax rounds only once the plain hold is comfortable, and only in the main block.
Re-test after three weeks. Expect the band not to move, because of the arithmetic above, and judge the work on how the position feels at the same depth rather than on the score alone. A reading older than 21 days is stale here, which is the app's own rule and a reasonable clock for anyone.
If the reason you are here is a chair, the evidence for that story is thinner than you have been told, and it is set out in tight hip flexors from sitting. If "tight" is the word you would use, tight hips: the causes separates the four things it can mean. The cluster is hips and legs.
Questions
How do I know the stretch is in my hip and not my back?
Set the pelvis before you move. Tuck the tailbone under, squeeze the back glute, keep the ribs down, and notice that tension appears across the front of the hip before you have travelled forward at all. If you can slide a long way forward with no sensation until the very end, the pelvis has tipped and the lower back is taking the range.
How long should I hold a hip flexor stretch?
Forty seconds is the app's default for the half-kneeling version and 30 for the standing one, which sits just above the 30 seconds that hamstring trials found sufficient. Doing it on more days matters more than holding it longer on one day. Total time across the week is the variable that the dose research keeps landing on.
Is the psoas the muscle I am stretching?
Partly, and you cannot tell from the inside. The hip flexor group includes the psoas, the iliacus and the rectus femoris, and a half-kneeling position loads all of them to some degree, with the rectus femoris taking more when the back knee is bent further. Naming one muscle sounds precise and adds nothing you can act on.
Why does my hip flexor score never change?
Because the test has three bands worth 25, 60 and 90, so the smallest change it can record is 30 points. Real improvement of a few degrees has nowhere to land on a scale that coarse. Judge the work by how the same depth feels and by whether the position is available on more days, not by a band that was never built to detect small movement.
One side reads a band lower. Is that an imbalance?
By the app's rule it is, because any one-band gap on this test is at least 30 points and the imbalance threshold is 12. That is a property of the instrument rather than a fact about you. Take the test again on a separate day with identical setup, and only treat it as real if it repeats.
Can I do the contract-relax version in a warm-up?
No. Contract-relax work belongs in the main block of a session, on a hip that is already warm, and the app will not schedule it in a warm-up. The sequence is a normal hold, a 6-second contraction at about 40% effort, a 2-second release, then 15 to 20 seconds deeper, for two or three rounds.