Elmsword
← Fitness

Fitness · Research

Resistance training and joint flexibility: a positive pooled effect against no exercise, but a prediction interval that crosses zero

A 2025 systematic review with a three-level meta-analysis pooled 36 studies, 1,469 participants and 169 effect sizes that compared resistance training with non-exercising control groups in healthy adults. Joint flexibility improved on average (g = 0.6325) and high-intensity protocols were linked to larger gains, but the 95% prediction interval ran from −0.3463 to 1.6113 and no included study was at low risk of bias.

The short version

This 2025 review in the Journal of Strength and Conditioning Research asked whether resistance training changes joint flexibility in healthy adults. It pooled 36 studies with 1,469 participants. Every study compared a resistance training group with a control group that didn't exercise. Stretching was not part of the comparison, so the review can't say whether lifting works as well as stretching.

On average, the training groups gained more flexibility than the controls. The pooled effect was g = 0.6325 (95% CI 0.4762 to 0.7888). Hedges' g, the effect size the paper uses, takes the change from before to after training in each group, divides it by the spread (standard deviation) of the starting values, and subtracts the control group's change from the training group's, with a small-sample correction. By the classic rule of thumb the paper cites, 0.2 is small, 0.5 medium and 0.8 large. The authors add that those cutoffs depend on context and that no such benchmarks exist yet for flexibility.

The catch is how much the studies disagreed. The 95% prediction interval ran from −0.3463 to 1.6113. A confidence interval says how precisely the average was estimated. A prediction interval says where the effect in a new, similar study setting would likely land once the differences between studies are counted. This one crosses zero, so some programs and groups may see no gain at all. The authors say that this spread "prevents us from concluding that RT has a positive effect on joint flexibility."

“However, the high overall risk of bias and substantial heterogeneity limit our ability to draw definitive conclusions.”

Favro et al. · Abstract · full text

What the researchers did

Stretching has a large research base. Lifting's effect on flexibility has much less, and an earlier review hadn't tested training variables such as intensity, volume or frequency as moderators.

“Despite the large body of evidence regarding the effectiveness of muscle stretching exercises, little is known about the effect of resistance training on flexibility.”

Favro et al. · Abstract · full text

The team registered the review in PROSPERO and searched 9 academic search tools. After duplicates were removed, 18,355 records were screened. To qualify, a study had to include healthy adults aged 18 or older, run a resistance training program lasting a minimum of 4 weeks, and measure flexibility or range of motion. Programs that mixed lifting with other training were left out, and so was Pilates.

Participants averaged 46.2 ± 23.1 years of age (from 35 of the 36 studies). Twenty-two studies used sedentary people, 7 used active people and 7 didn't say. Programs lasted 11.08 ± 6.16 weeks on average (range 4–24), averaging 2.93 sessions per week and 20.27 sets per session. Most studies measured flexibility with the sit-and-reach test or a modified version (21 of 36). Twelve used at least one goniometric measurement of a joint's range of motion.

Because most studies reported several flexibility outcomes, the authors used a three-level meta-analysis that keeps outcomes from the same study grouped together. That gave 169 effect sizes. They then ran meta-regressions, testing one moderator at a time, to see whether things like intensity, age or activity level were linked to bigger or smaller effects. Strength was analyzed separately as a secondary check that the programs worked.

Key findings

Resistance training vs non-exercising control: joint flexibility and strength (effect sizes are Hedges' g; positive = the training group improved more)
MeasureEstimate95% CILocation
Flexibility, all 36 studies (169 effect sizes)g = 0.63250.4762 to 0.7888Results
95% prediction interval for that estimate–PI −0.3463 to 1.6113Results; Discussion
Heterogeneity (I²)69.3% total; 38.5% between studies–Results
Flexibility without the 4 effect sizes above 3g = 0.61780.4636 to 0.7721Results
Low-intensity protocols *g = 0.2843−0.0780 to 0.6466 (PI −0.6936 to 1.2622)Results; Figure 3
High-intensity protocols *g = 0.74950.4960 to 1.0030 (PI −0.1935 to 1.6925)Results; Figure 3
Activity level (moderator)sedentary 0.7009, active 0.4530; p = 0.2543, not significant–Results
Age (moderator)p = 0.7570, not significant–Results
Strength, secondary outcome (94 outcomes, 26 studies) †g = 1.10560.7220 to 1.4892Results; Figure 7
Risk of bias, randomized studies (RoB 2)64.6% of outcomes “some concerns”, 35.4% “high”–Results; Figure 2
Risk of bias, non-randomized studies (ROBINS-I)all 9 “serious” or worse; 3 “critical”–Results
Source: Favro et al. 2025, Results and Discussion. * Intensity was tested as a moderator across studies, based on 129 effect sizes; high vs low p < 0.0225. It is an association, not a direct comparison inside one trial. † The strength funnel plot showed marked asymmetry, compatible with a small-study effect. CI = confidence interval · PI = prediction interval · Full text (PMC)
Dot-and-interval chart of the joint flexibility effect of resistance training versus non-exercising control, in Hedges' g. All 169 effect sizes: 0.6325, 95% CI 0.4762 to 0.7888, 95% prediction interval −0.3463 to 1.6113. Without the 4 effect sizes above 3: 0.6178, 95% CI 0.4636 to 0.7721, prediction interval not reported. Low-intensity protocols: 0.2843, 95% CI −0.0780 to 0.6466, prediction interval −0.6936 to 1.2622. High-intensity protocols: 0.7495, 95% CI 0.4960 to 1.0030, prediction interval −0.1935 to 1.6925. Every prediction interval crosses zero.
Source: Results section and Figures 3 and 4 of Favro et al. Dots are estimates, thick lines are 95% CIs and dashed lines are 95% prediction intervals. Right of 0 means the training groups gained more flexibility than the controls. The paper's text doesn't give values for moderate-intensity protocols or a prediction interval for the model without the 4 largest effect sizes, so neither is drawn.

How to read the result

The average points the right way, and it held up when the authors removed the 4 effect sizes above 3 (g = 0.6178, 95% CI 0.4636 to 0.7721). One of those, a trunk extension measurement, had an effect size of 15.4, and the authors couldn't confirm it with the original study's team.

High-intensity protocols were linked to a larger effect (g = 0.7495) than low-intensity protocols (g = 0.2843), and the low-intensity confidence interval included zero. This is an association across studies from a meta-regression, not a head-to-head comparison, and only 12 studies reported intensity as a percentage of one-repetition maximum (%1RM). The authors suggest future studies test low against high intensity directly.

Rest time between sets also came out as a significant moderator (p = 0.0007). The authors think it mostly tracks intensity, so it isn't a separate finding.

“Another significant moderator was rest time between sets, although higher rest times are usually used with higher intensities, which would mean that the variable is a proxy for exercise intensity.”

Favro et al. · Discussion · full text

Neither activity level nor age was a significant moderator. Sedentary groups had a higher estimate (0.7009) than active groups (0.4530), but p was 0.2543.

Strength rose a lot more than flexibility (g = 1.1056, 95% CI 0.7220 to 1.4892), which suggests the programs worked as strength training. That number comes with its own warning: the strength funnel plot was lopsided in a way that fits small studies reporting bigger effects.

What this doesn't tell us

Whether lifting can stand in for stretching. Every study compared resistance training with doing nothing. The review didn't test stretching or combined programs, and the authors say the results shouldn't be used as a reason to drop stretching.

“Still, the results of this work should not be interpreted as a basis to discard stretching exercise routines, which are still important as a safe and easy to administer complement in training routines for healthy persons.”

Favro et al. · Practical Applications · full text

What happens in people who already train. Only 7 studies used active participants, and only one (Junior et al.) explicitly included people with strength training experience.

“Future studies conducted on active persons may be warranted to confirm these effects.”

Favro et al. · Discussion · full text

How solid the evidence is. No included study was at low risk of bias. Only one study had a pre-prepared analysis plan, and only 4 described their randomization in meaningful detail. Results were analyzed only for people who finished the programs, with no intention-to-treat analysis, and overall dropout was about 6% (93 of 1,469).

What exactly improved. Most studies used sit-and-reach, a test that involves several joints at once. The authors also point out that many different flexibility tests were used, which may have added to the disagreement between studies.

A few loose ends sit inside the paper itself. The model's residuals weren't normally distributed. The sex moderator is reported inconsistently: the abstract gives p = 0.0429 and the Results text gives p = 0.496, so this article doesn't treat it as a finding. The full study table and full meta-regression results are in supplementary files that this article didn't review.

How to apply it

The authors present resistance training as a possible way to improve joint flexibility, with stretching kept as a complement. They don't give a specific program.

Opinion, not a finding from the paper: If I were in your spot, I would count any flexibility gain as a possible side benefit of a lifting program, and I would keep stretching if it's already in your routine, because this review never compared the two. For general guidance on setting up resistance training, see the site's article on the ACSM hypertrophy principles.

A safety note: if you're new to lifting, start lighter and learn the form of each exercise before adding load. This article isn't injury or rehab advice. If you have pain, an injury or a health condition, talk to a qualified professional.

Sources

Source: Favro et al., Journal of Strength and Conditioning Research 2025; 36 studies, 1,469 participants, 169 effect sizes. DOI 10.1519/JSC.0000000000005000.

  1. StudyFavro F, Roma E, Gobbo S, Bullo V, Di Blasio A, Cugusi L, Bergamin M. The influence of resistance training on joint flexibility in healthy adults: a systematic review, meta-analysis, and meta-regression. Journal of Strength and Conditioning Research. 2025;39(3):386–397.
    DOI: 10.1519/JSC.0000000000005000 · Full text (PMC, CC BY 4.0)

This is a research explainer for general education, not medical advice.