The "corrective exercise" myth: Stop trying to fix problems that aren't broken

A trainer watches you do a bodyweight squat, frowns, and delivers the verdict: your left knee caves in, your right hip is "locked up," your glutes aren't "firing." You are, it turns out, a collection of dysfunctions. The good news is that a corrective program can fix you before all of this catches up and injures you. The bad news, and it is worth saying plainly, is that almost none of that story survives contact with the evidence.

The corrective-exercise model rests on a chain of assumptions: that there is a "correct" way to move and stand, that a screen can spot deviations from it, that those deviations predict pain or injury, and that specific exercises can reverse them. Each link in that chain has been tested. Most of them have failed.

The flaw is in the first premise: what is "correct"?

The word "corrective" smuggles in a claim before anyone has thrown a plate on a bar: that there is an incorrect state to be corrected against a known correct one. But as the clinicians behind the "corrective exercise trap" argument point out, humans vary enormously in posture, joint angles, and movement, and much of that variation is normal, harmless, and task-dependent. We move differently to accomplish different things. Flagging a small deviation as a "dysfunction" assumes a precision, a single right answer, that the anatomy simply does not support.

So before we even ask whether corrective exercise works, there is a prior question the model rarely answers: correct according to whom, and measured against what validated standard? In most cases, there isn't one. The "ideal" being enforced is often a textbook diagram, not a research-derived threshold that separates the injured from the uninjured.

Screening can't tell who is going to get hurt

In 2016, Roald Bahr published a critical review in the British Journal of Sports Medicine with a title that leaves little ambiguity: "Why screening tests to predict injury do not work, and probably never will." To validate a screening test you need a strong prospective association between the test result and injury, then adequate test properties (sensitivity and specificity) in a real population. His conclusion was that there is currently no example of a sports-injury screening test that clears that bar.

The reason is intuitive once you see it. Take Bahr's own data: 614 professional soccer players in Qatar, tested for hamstring and quadriceps strength, then tracked for four seasons during which 190 suffered hamstring strains. When you try to set any cutoff to separate the "weak, at-risk" players from the safe ones, you drown in false positives (weak players who never got hurt) and false negatives (strong players who did). A test with a statistically significant association can still be, as Bahr put it, no better than a coin toss at the individual level. Association is not prediction.

The Functional Movement Screen, tested directly

The FMS is the most commercially successful movement screen in the world, so it has been studied more than any other. In 2017, a systematic review with meta-analysis in the BJSM pooled 24 prospective cohort studies. Their conclusion was blunt: the strength of association between FMS composite scores and subsequent injury does not support its use as an injury-prediction tool.

This is not a case of a screen that needs more research. The FMS actually has good inter-rater reliability, meaning two coaches will score the same person similarly. What it lacks is validity: a reliable score that predicts nothing is a precise measurement of the wrong thing. As one of the review's authors summarized, any way they sliced the data, the screen was no help in identifying who would get injured.

The "structural damage" on your scan is probably normal

A close cousin of corrective exercise is the belief that a specific structural finding, a disc bulge, "degeneration," an asymmetry, is the cause of your pain and must be addressed. The problem is that when researchers scan people with no pain at all, they find these "abnormalities" everywhere.

Brinjikji and colleagues pooled imaging from 3,110 pain-free individuals. Among symptom-free 20-year-olds, 37 percent already showed disc degeneration and 30 percent had a disc bulge. By age 80, those figures were 96% and 84%. Many of these findings are simply part of normal aging, like wrinkles on the inside, and must be interpreted in the context of the person, not treated as an automatic pain generator.

AN HONEST CAVEAT: This does not mean structure never matters. The same research group found in a separate meta-analysis that degenerative findings are somewhat more common in adults under 50 with back pain than in those without. The accurate statement is not "your scan is meaningless," it is "your scan cannot be assumed to explain your pain, because the same finding sits silently in millions of people who feel fine." Structure is one input, not a verdict.

"Sit up straight" is not the injury-prevention advice you were told

Posture correction is corrective exercise's most mainstream expression, embedded in ergonomic assessments, "text neck" panic, and endless glute-activation drills. Yet a 2019 editorial in the Journal of Orthopaedic & Sports Physical Therapy, titled "Sit Up Straight: Time to Re-evaluate," states there is no evidence to support posture or movement screening for the primary prevention of workplace pain. People come in different shapes with naturally varying spinal curves, and there is no single alignment that has been shown to protect against pain.

There is a subtler harm here worth naming. When a clinician or trainer tells someone their posture is "faulty" and links it to their pain, and then the pain persists despite the corrective drills, the message that lands is that the body is fragile and broken. The JOSPT authors flag this directly: appearance-focused, fear-based language can encourage hypervigilance and worsen outcomes. Telling a healthy person that ordinary sitting is damaging them is not neutral advice; it can be iatrogenic, meaning the intervention itself causes harm.

So what actually works? Capacity, not correction

Here is the pivot the outrage-free version of this argument insists on: rejecting corrective exercise is not an argument for doing nothing, and it is not a claim that exercise doesn't prevent injury. Exercise absolutely does. It just isn't the corrective, fix-the-flaw kind. The most useful meta-analysis here is Lauersen 2014, which pooled 25 randomized controlled trials covering more than 26,000 people and 3,464 injuries, and stratified by intervention type.

The results are clarifying. Stretching returned a relative risk of 0.963, which is statistically no effect at all. Proprioception training helped (RR 0.55). Strength training was the standout, cutting injuries to under a third of the control rate (RR 0.315), and roughly halving overuse injuries. A 2018 follow-up from the same group confirmed strength training as the superior, dose-dependent, and safe preventer: more of it, within reason, worked better.

Notice what these interventions have in common, and what they lack. None of them require a screen. None of them diagnose a "dysfunction." They build tissue tolerance and coordination across theboard, so the body can handle more load with a wider margin before something fails. That is a fundamentally different model from hunting for the one broken part and fixing it. You are not repairing a fault; you are raising a ceiling.

What the Evidence Support

The corrective-exercise myth is not that exercise is useless, it is that the diagnostic story wrapped around it, find the flaw, screen for the dysfunction, correct the imbalance, is built on premises the research does not support.

• Strong: screening tests, including the FMS, do not predict individual injury. A group-level association is not the same as a usable prediction.
• Strong: structural findings like disc bulges are common in pain-free people, so a scan cannot be assumed to explain symptoms, though it is not always irrelevant either.
• Strong: there is no evidence that posture screening prevents pain, and fear-based "faulty posture" messaging can backfire.
• Strong: strength training meaningfully reduces injuries; general stretching does not. Capacity beats correction.
• The nuance corrective advocates get right: individual assessment matters for people who are actually injured, and movement screens can be fine as benchmarks of function as long as no one pretends they predict the future.

The most freeing thing is also the simplest. Your body is not a fragile machine one asymmetry away from breaking down, and you do not need to be diagnosed and corrected before you are allowed to train hard. For the overwhelming majority of people, the answer to "what should I fix first?" is: nothing. Get stronger, move often, load progressively, and let a robust body do what robust bodies do, which is tolerate a great deal without being optimized into fragility first.