Does stretching prevent injury is the question this whole category is built on, and the honest answer is no. Pooled randomised trials found stretching had no significant protective effect against sports injury. The same pooled analysis found strength training reduced injuries to under a third. A separate trial randomised 1,538 army recruits to stretch or not stretch before exercise and found no reduction in injury at all.
That is the finding, stated first, because it decides what everything below is worth. This page sets out the trials, what they actually measured, the honest limits of the evidence, and what a mobility app is entitled to claim once it has read them.
The short answer
- Stretching has not been shown to prevent injuries. Multiple randomised trials, two decades, one direction.
- Strength training has. To under a third for injuries overall, about half for overuse.
- The largest single trial randomised 1,538 army recruits and found nothing.
- For hamstrings specifically, the prospective risk factors were architecture and strength, not flexibility.
- Screening for who will get hurt does not work either.
- Limber does not claim injury prevention anywhere in the app or on this site.
Does stretching prevent injury
Lauersen, Bertelsen and Andersen published the meta-analysis in the British Journal of Sports Medicine in 2014. They pooled randomised controlled trials of exercise interventions for preventing sports injuries and separated the interventions by type — strength training, proprioception work, stretching, multi-component programmes.
Strength training reduced sports injuries to under a third and overuse injuries to about half. Stretching showed no significant protective effect. Proprioception and multi-component programmes came somewhere between.
That is a comparison, not a hedge. Both interventions were tested by the same method, in the same analysis, against the same outcome. One of them worked and one did not.
Grade: reasonably solid. A meta-analysis of randomised trials is the best design this question is ever likely to get, and the effect size for strength training is large enough that it does not depend on one included study.
The 1,538 recruits
Pope and colleagues ran the trial that people usually mean when they say the evidence is against stretching. Published in Medicine & Science in Sports & Exercise in 2000, it randomised 1,538 army recruits to a stretching protocol before exercise or to no stretching, through twelve weeks of basic training, and counted injuries.
There was no reduction in injury.
Herbert and Gabriel reviewed stretching before and after exercise in the BMJ in 2002 and reached the same conclusion, for injury risk and for soreness together.
Grade: solid for the specific intervention tested. Large, randomised, prospective, on a population that gets injured often enough to give the study power. It is the kind of trial that almost never gets run in this field, and it came back empty.
Hamstrings, prospectively
The specific case is worse for flexibility than the general one.
Timmins and colleagues followed elite footballers prospectively in the British Journal of Sports Medicine in 2016 and looked for the modifiable factors that predicted hamstring injury. They found short biceps femoris fascicles and weak eccentric knee flexors. How far the hamstring stretched did not appear as a modifiable risk factor.
Grade: reasonably solid, in one population. Elite male footballers are not everyone, and a prospective cohort is not a trial. But it is the best-designed look anyone has taken at the single injury stretching is most often sold against, and it pointed at architecture and strength.
The consistent picture across all three: the thing that reduces injuries is making tissue stronger, not making it longer. Hamstring stretches prints the same finding on the page that teaches the stretch.
What the trials did not test
Now the caveats, which come after the finding and do not cancel it.
They tested stretching around a session. Almost all of this literature is about stretching immediately before, or immediately after, exercise. Nobody has taken a group of people, put them through months of sustained range work as a programme in its own right, and counted their injuries against a control group. That trial does not exist. So "stretching before you train does not protect you" is established; "range work over a year has no protective value" is untested rather than refuted.
"Injury" is a crude outcome. These analyses pool a hamstring tear with shin pain and an ankle sprain. Real effects on one injury type can vanish inside a total that is dominated by another.
The dose may have been wrong. Most protocols were short holds, a few minutes, immediately pre-exercise. The dose that raises range in the chronic trials — 30 seconds a hold, most days, for weeks — is a different intervention.
None of that is a reason to believe stretching prevents injury. It is a reason to say the honest thing: the version that has been tested does not work, and the version people mean when they argue back has not been tested.
Screening does not rescue it
The usual reply is that the trials failed because they stretched everyone rather than the people who needed it. Screen first, then intervene.
That does not survive contact with the screening literature either. Moran and colleagues' systematic review in the British Journal of Sports Medicine in 2017 found Functional Movement Screen composite scores predict injury poorly. Bahr argued in the same journal in 2016 that this is a general property of screening rather than a fault in one test: to identify the individuals who will get hurt, a test needs a discriminative power that essentially no movement screen has demonstrated.
Grade: solid, and it runs against the way screens are sold.
The arithmetic behind that is worth understanding, because it is not a criticism of any particular test. Injuries are uncommon events. A screen with what sounds like good accuracy, applied to a population where only a small fraction will get hurt, still labels far more healthy people as at risk than it correctly identifies. You end up intervening mostly on people who were never going to be injured, which is how a test with a real association attached to it produces no useful prediction for any individual. That is why Limber's readings are framed as a record of where you are rather than a forecast of what will happen.
There are genuine associations in the literature — Backman and Danielson found low ankle dorsiflexion associated with developing patellar tendinopathy in junior basketball players in the American Journal of Sports Medicine in 2011, and Malliaras, Cook and Kent reported the same association in volleyball in the Journal of Science and Medicine in Sport in 2006. Associations are not interventions. No trial has raised a restricted range and shown fewer injuries afterwards.
What Limber will not claim
This site is written by people who sell a mobility app, so the position needs to be explicit.
Limber does not claim to prevent injury. Not in the app, not in a notification, not on the price screen, not here. There is no test in the battery that predicts an injury and no routine that reduces one, and the Range Score has never been compared against any population or any outcome. It is a reading of where your joints stop today.
Limber has no pain protocols and no sport packs. The only pain-related mechanism in the engine is an exclusion rule: movements you have flagged are removed from the pool before a routine is built rather than ranked lower inside it.
What the app does is measure range and plan range work. The Range Score composes your readings into one number across four of five stations and leaves the fifth blank, because the test that would feed it is not built. That gap is printed rather than filled.
If something hurts, that is a matter for a clinician who can examine you, not for a routine — when to see a physio sets out the thresholds. Nothing here diagnoses anything.
Where that leaves stretching
Somewhere honest, and smaller than the category would like.
Stretch because you want the range. Because a position — a deep squat, a hand behind your back, a fold to the floor — is somewhere you would like to be able to get to. Because it is a pleasant twenty minutes and the end of a hard day. Those reasons are enough and they do not need a trial to prop them up.
Then do the loading. It is the intervention with the prevention evidence behind it, and even a very small dose has shown clinical effects: Andersen and colleagues found in Pain in 2011 that two minutes a day of progressive resistance training reduced office neck and shoulder pain over ten weeks.
Nuzzo went further in Sports Medicine in 2020, arguing in The Case for Retiring Flexibility as a Major Component of Physical Fitness that flexibility's status rests on weaker justification than strength or aerobic capacity. Grade: a serious minority position, well argued — and worth reading even if you intend to keep stretching, which we do.
Mobility for runners and warm-up versus stretching apply this to the two places it bites hardest, and the whole cluster sits under mobility for training.
Questions
Does stretching before exercise prevent injuries?
No. The largest randomised trial put 1,538 army recruits through twelve weeks of basic training with or without pre-exercise stretching and found no reduction in injury, and the pooled analysis of injury-prevention trials found stretching had no significant protective effect. This is one of the better-tested claims in sports medicine and it did not hold up.
What does prevent injuries, then?
Strength training, by a wide margin. The same 2014 meta-analysis that found nothing for stretching found strength training reduced sports injuries to under a third and overuse injuries to about half. Proprioception and multi-component programmes came somewhere in between, and stretching came last.
Does stretching prevent muscle strains specifically?
Not on the evidence there is. Prospective work in elite footballers found the modifiable hamstring-injury risk factors were short biceps femoris fascicles and weak eccentric knee flexors rather than flexibility, which points at loaded eccentric work instead. Flexibility did not appear as the lever people assume it is.
Is there any version of this that might still work?
Possibly, and it has not been tested. Almost every trial studied stretching immediately around a session rather than months of sustained range work as a programme of its own, so the honest statement is that the tested version fails and the argued-about version is unstudied. That is a gap in the literature, not a hidden positive result.
Should I screen myself for injury risk first?
It will not tell you what you want to know. Movement-screen composites predict injury poorly, and the broader argument is that screening for individual risk needs a discriminative power no such test has shown. Readings are useful for tracking your own change over time and not for forecasting what will happen to you.
Does Limber claim its plans keep you healthy?
No. The app measures range and builds range work, and it makes no injury-prevention claim in any screen, notification or page. The Range Score has never been compared against a population or an outcome, and it is presented as a reading of where you are rather than a forecast. Anything that hurts belongs with a clinician who can examine you.