Core strength for back pain is one of the two answers the internet gives, and stretching is the other, and both camps claim more than their evidence supports. This page gives each side its best case at full strength, then shows what happened when the specific claims were tested, and ends with the finding that beat both of them — which is neither exciting nor easy to sell.
The short answer
- The core-stability story began with a real laboratory finding and outgrew it badly.
- Trunk-control programmes have not beaten other exercise in randomised review.
- Trunk muscle endurance relates to back pain better than any range or posture measure.
- Stretching helps chronic back pain modestly, and no more than other movement does.
- Exercise beats no exercise. Which kind matters much less than either camp says.
Read this first
Get examined rather than starting a programme if you have loss of bladder or bowel control, difficulty passing urine, or numbness in the saddle area alongside back pain — that needs assessment the same day. Also: progressive leg weakness, back pain after significant trauma, back pain with fever or unexplained weight loss, a history of cancer, or pain that is constant and worse at night. Symptoms travelling below the knee deserve assessment before any routine. When to see a physio covers the ordinary version of that decision.
Nothing here diagnoses anything and nothing here is treatment.
Where the core story came from
It began with a genuine and interesting finding.
Hodges and Richardson, in Spine in 1996, measured the timing of trunk muscle activation during rapid arm movements. In people without back pain, transversus abdominis fired before the limb moved — an anticipatory contraction. In people with back pain, that onset was delayed.
That is a real result and it was reported carefully. What happened next was not careful. A delayed onset in a laboratory became "a weak core causes back pain", which became a training industry, which became the near-universal advice to draw the navel toward the spine before doing anything at all.
Two things went wrong in that translation. A difference between groups was read as a cause, which the design cannot establish — a delayed onset is at least as plausible as a consequence of pain as a cause of it. And a millisecond timing difference was read as a strength deficit, which is a different variable entirely.
Lederman, in the Journal of Bodywork and Movement Therapies in 2010, set out the case against the whole edifice under the title "The myth of core stability", arguing that the specific mechanisms claimed for core training had not been demonstrated and that the exercises were not doing what they were said to do.
Grade for the original finding: solid but narrow. Grade for what was built on it: thin.
Core strength for back pain
What happened when the training was actually tested?
Saragiotto and colleagues, Cochrane, 2016. Motor control exercise for chronic low back pain — the trunk-control programmes, reviewed properly. It produced benefits, and it came out similar to other forms of exercise rather than superior. That is the review the whole field waited for and it did not deliver a winner.
Hayden and colleagues, Cochrane, 2021, on exercise therapy for chronic low back pain more broadly: exercise probably reduces pain and improves function modestly against no treatment or usual care, and the comparisons between exercise types produce small differences that do not consistently favour anything.
Grade: moderate, and consistent in its lack of a winner. Large pooled literatures, real but modest effects, and repeated failure to separate the branded approaches from each other.
There is one result that goes the strength camp's way and it is not about abdominal drills.
Biering-Sorensen, in Spine in 1984, followed a large working population for a year and found that poor isometric endurance of the back extensors predicted first-time low back pain in men. Prospective, which is stronger than the cross-sectional work most of this field runs on.
Nourbakhsh and Arab, in JOSPT in 2002, looked at 600 people and found trunk muscle endurance strongly related to low back pain, while lumbar lordosis, abdominal muscle length, and hip flexor and hamstring length were not related at all.
Notice what those two findings are about. Endurance of the back extensors — how long you can hold a position — rather than the strength of an abdominal muscle or the timing of its onset. The evidence that survives points at capacity over time, which is not what the core-stability industry sells.
The flexibility side
The stretching camp's case is thinner than it thinks and better than nothing.
Sherman and colleagues, Archives of Internal Medicine, 2011. Twelve weeks of stretching classes matched twelve weeks of yoga classes for function in chronic low back pain, and both beat a self-care book, in 228 randomised adults. A real result and a real active comparator.
Against it, three findings the camp does not quote.
Nourbakhsh and Arab found no relationship between low back pain and hamstring or hip flexor length in their 600 people. That is the specific claim under most back-stretching advice, tested, and unsupported.
Katalinic and colleagues' Cochrane review of stretch reported high-quality evidence that stretch produces no clinically important change in joint mobility. The honest caveat is that it studied clinical populations at risk of contracture rather than healthy adults chasing range, so it does not transfer directly. It is still a Cochrane review of stretching concluding the intervention did not do its job.
And Stanton and colleagues, in Scientific Reports in 2017, found that felt stiffness in people with back pain did not match objectively measured spinal stiffness. If the tightness is a perception rather than a mechanical property, stretching it will feel good and change nothing. Is your back tight or weak works that through.
Grade: moderate for stretching as movement, thin for stretching as a mechanism.
What both camps overstate
Set side by side, the two positions fail in the same way.
The core camp claims a mechanism — instability, a lazy deep muscle, a timing fault — that the trials have not confirmed and that its own founding study was too small a design to establish.
The stretching camp claims a mechanism — shortness, adhesion, a region needing lengthening — that the measurement studies have not confirmed either, and that Stanton's work directly undercuts.
And both are outperformed by the same unglamorous finding: exercise of almost any kind modestly helps, and the kind matters much less than the doing. Steffens and colleagues, in JAMA Internal Medicine in 2016, found exercise reduced the risk of a future episode where education alone, back belts and shoe insoles did not. Cochrane finds exercise beats no exercise and cannot find a winner among the types.
So the practical answer to the question in this page's title is: do both, or do whichever you will still be doing in three months, and stop looking for the one that fixes the mechanism, because nobody has identified the mechanism.
If you want a reading to work from, the deep squat test reads the whole lower body in four bands over a 30-second hold, and the toe touch test reads the posterior chain in five — with the warning that a toe touch is a loaded forward fold and is the wrong test on a day when bending forward provokes symptoms. Between them they are the entire lower-back reading in Limber, at half weight each, both labelled as proxies. Neither of them measures strength at all. The app has no strength measure of any kind, which is worth saying on a page about whether you are weak. The positions themselves are in lower back stretches and the cluster sits under lower back.
Questions
Is core strength or flexibility better for back pain?
Neither has beaten the other, and both have failed to beat generic exercise. Cochrane's 2016 review found motor control exercise similar to other forms rather than superior, and its 2021 review of exercise for chronic low back pain found modest benefits with small differences between types. The best-supported statement is that exercise beats no exercise and adherence decides the rest.
Does a weak core cause back pain?
That claim outran the study it came from. Hodges and Richardson found a delayed onset of a deep abdominal muscle in people with back pain in 1996, which is a difference between groups rather than a demonstrated cause, and a timing difference rather than a strength deficit. What has held up better is trunk muscle endurance — how long you can hold a position — which related strongly to back pain in a study of 600 people and predicted first episodes in a prospective cohort.
Should I do planks for my back?
If you enjoy them and will keep doing them, they are a reasonable way to train trunk endurance, which is the variable with the best support in this area. What is not supported is the idea that a specific abdominal drill corrects a mechanism, or that one exercise is the answer. Persistent back pain is a matter for a clinician who can examine you rather than for a plank.
Are tight hamstrings causing my back pain?
There is no good evidence that they are. In a study of 600 people, low back pain was not related to hamstring length, hip flexor length, abdominal muscle length or lumbar lordosis, while trunk muscle endurance was strongly related. That does not make hamstring work pointless, and it does make "your tight hamstrings are pulling on your pelvis" a claim with very little behind it.
Should I strengthen first or stretch first?
The evidence does not order them, because it cannot separate them. What it does say is that exercise reduces the risk of a future episode where education, back belts and insoles do not, and that no type of exercise clearly beats another. Start with whichever you will do three times a week without negotiating with yourself about it.
Can I measure whether my core is weak?
Not with anything on this site. Limber measures range and has no strength measure of any kind — its lower-back number is half a forward fold and half a squat, both labelled as proxies, and neither reads a muscle's capacity. Trunk endurance is measurable with timed hold tests in a clinic, and if the answer matters to your plan a professional can run one in a few minutes.