Stretching is the one injury-prevention measure the trials did not support

Stretching does not appear to prevent sports injuries. In the meta-analysis that separated injury-prevention programmes by what they actually contained, stretching was the only category that showed no effect — a relative risk of 0.963, with a confidence interval from 0.846 to 1.095 that comfortably includes no difference at all [s1]. Strength training, in the same analysis, was associated with a relative risk of 0.315 [s1].

That is a striking gap, and it is worth being careful about what produced it, because the two estimates do not rest on equally solid ground.

What the analysis pooled

Researchers publishing in the British Journal of Sports Medicine in 2013 searched PubMed, EMBASE, Web of Science and SPORTDiscus, screened 3,462 records, and analysed 25 randomised controlled trials covering 26,610 participants and 3,464 injuries [s1]. Twelve of the trials had not accounted for clustering effects and were adjusted before pooling [s1].

The overall effect estimate across all trials was heterogeneous — which is the reason the stratified analysis exists. Split by what the intervention actually was, the pattern was consistent: strength training relative risk 0.315 (0.207–0.480), proprioception training 0.550 (0.347–0.869), programmes combining multiple exposures 0.655 (0.520–0.826), and stretching 0.963 (0.846–1.095) [s1]. Both acute injuries (RR 0.647, 0.502–0.836) and overuse injuries (RR 0.527, 0.373–0.746) were reduced by physical activity programmes overall [s1]. Intention-to-treat sensitivity analyses produced more robust estimates rather than weaker ones [s1].

The authors' summary is unusually direct: consistently favourable estimates were obtained for all injury prevention measures except stretching [s1].

The strength-training figure is smaller than it looks

A follow-up review by the same lead author, published in 2018, went looking specifically at randomised trials of strength training as primary injury prevention [s2]. It found six. Those six studies tested five different interventions against four distinct outcomes, in 7,738 participants aged 12 to 40, who sustained 177 acute or overuse injuries between them [s2]. Five were conducted in Europe and one in Africa, published between 2003 and 2016 [s2].

The cluster-adjusted intention-to-treat estimate was a relative risk of 0.338 (0.238–0.480), consistent across robustness tests, with no publication bias detected on formal testing [s2]. A meta-regression found a dose relationship: a 10% increase in strength training volume was associated with a reduction in injury risk of more than four percentage points [s2].

That is a large effect, graded by the authors as high strength of evidence [s2]. But 177 injuries across six trials is a thin foundation for a claim of this size, and the trials differed considerably in population and intervention. The honest statement is that strength training has the best evidence of anything tested — not that the size of its benefit is precisely known.

Stretching does not do the other thing it is credited with either

The second common justification for stretching is that it reduces next-day soreness. A Cochrane review of randomised and quasi-randomised trials, updated in 2011, tested that directly across 12 studies [s3]. One was a large field trial with 2,377 participants, 1,220 of whom were allocated to stretching; the other 11 were small, with between 10 and 30 participants in the stretch condition [s3]. All studies carried moderate or high risk of bias, and the quality of evidence was rated low to moderate [s3].

Results were consistent. Pre-exercise stretching reduced soreness one day later by an average of half a point on a 100-point scale (mean difference −0.52, 95% CI −11.30 to 10.26, three studies) [s3]. Post-exercise stretching reduced it by about one point (−1.04, 95% CI −6.88 to 4.79, four studies) [s3]. The single large trial found that stretching both before and after exercise reduced peak soreness over a week by an average of four points on the same scale (−3.80, 95% CI −5.17 to −2.43) — statistically significant, and, as the reviewers put it, very small [s3]. Their conclusion was that stretching, whenever performed, does not produce clinically important reductions in delayed-onset muscle soreness in healthy adults [s3].

What the evidence does not say

None of this is evidence that stretching is harmful, and none of it addresses the reasons people stretch that have nothing to do with injury statistics: range of motion, the requirements of a specific sport, or simply that it feels good. Nor does it separate types of stretching cleanly — the 2013 analysis pooled stretching as a category, and the underlying trials were not designed to compare static against dynamic protocols on injury outcomes.

What it does say is that the specific promise attached to stretching in most gym advice — that it lowers your chance of getting hurt — is not visible in randomised trials, while the same trials show a substantial signal for progressive strength work. Whether an athlete or a recreational exerciser should change anything on that basis is a question for them and their clinician or coach, and one that depends on the sport, the injury history and the time available.

This article is informational and is not medical advice.

Sources

• The effectiveness of exercise interventions to prevent sports injuries: a systematic review and meta-analysis of randomised controlled trials — British Journal of Sports Medicine, 2013-10-07
• Strength training as superior, dose-dependent and safe prevention of acute and overuse sports injuries: a systematic review, qualitative analysis and meta-analysis — British Journal of Sports Medicine, 2018-08-21
• Stretching to prevent or reduce muscle soreness after exercise — Cochrane Database of Systematic Reviews, 2011-07-06