Creatine and Resistance Training: A Comprehensive Meta-Analysis Confirming Established Muscle Benefits

The combination of creatine supplementation and resistance training is arguably the most well-documented synergistic intervention in sports nutrition. While its efficacy in improving strength and muscle mass is widely accepted, researchers continue to refine our understanding of how individual characteristics—like age or training experience—might influence these outcomes. A recent meta-analysis sought to synthesize these effects using modern statistical techniques to account for multidimensional variables across decades of research.

Key Takeaways

• Strength Gains: Creatine combined with resistance training produced a moderate, statistically significant increase in both upper- and lower-body strength compared to training alone, though signs of publication bias suggest the true effect is roughly a third smaller than the headline figure.
• Lean Mass Increases: The intervention yields consistent improvements in lean body mass, which accounts for most of the overall increases in total body mass.
• No Significant Change in Fat Mass: Creatine did not significantly reduce fat mass, though because lean mass rose, body-fat percentage fell slightly.
• Age Signal Is Uncertain: Exploratory analyses linked older age with smaller strength gains, but adults aged 60 and over still gained nearly as much as young adults; the age difference came mainly from a small middle-aged group.
• Low-to-Moderate Certainty: The authors rated the certainty of the evidence as low to moderate, and the underlying trials were small, short, and mostly in men.

The Study Design

This systematic review and meta-analysis evaluated the effects of resistance training combined with creatine supplementation across 63 included studies, capturing a total sample size of 1,489 participants. The researchers examined outcomes spanning muscle strength (both upper- and lower-body) and body composition metrics, including lean mass, fat mass, body-fat percentage, and total body mass.

To manage the complexity of multiple effect sizes extracted from the same trials, the authors utilized a three-level random-effects model¹. They also conducted exploratory subgroup analyses and meta-regression² to investigate whether participant characteristics (age, gender, training experience) or intervention variables (duration, weekly frequency, daily dose, dosing strategy, and whether a loading phase was used) moderated the results.

The included trials were small and short: each enrolled between 8 and 45 participants and lasted 4 to 12 weeks, with 3 to 5 training sessions per week, and all used creatine monohydrate. Participants were aged 18 to 75; 970 were men, 248 were women, and 271 were not reported by sex. About half (49%) were untrained, 28% were resistance-trained, and 23% were athletes. The review protocol was prospectively registered in PROSPERO³ (CRD420251078957), and two reviewers independently assessed each trial's risk of bias and graded the certainty of the evidence using GRADE⁴.

The Findings

The results line up with the established scientific consensus. Compared to a resistance training protocol without creatine, the addition of creatine supplementation resulted in significant, moderate effect sizes for upper-body strength (Hedges' g = 0.54, p < 0.001) and lower-body strength (g = 0.55, p < 0.001, after excluding one outlier study).

Two caveats temper those strength figures. First, statistical tests found significant signs of publication bias⁵ in both strength analyses. When the authors adjusted for the studies that appear to be missing, the effects fell to g = 0.36 for upper-body strength and g = 0.35 for lower-body strength: still significant, but about a third smaller. Second, strength results varied between studies enough that the range expected for a future study (the 95% prediction interval) included zero for both outcomes.

Body composition also shifted favorably. Lean mass saw a significant increase (g = 0.30, p < 0.001), driving a modest rise in total body mass (g = 0.12, p = 0.04). Notably, the intervention had no significant effect on absolute fat mass (g = -0.10, p = 0.14). However, because lean mass increased while fat mass held steady, overall body-fat percentage experienced a slight, significant reduction (g = -0.14, p = 0.02).

Age was the only factor that significantly changed the strength results, and the pattern is less simple than "younger is better." Young adults (18–44) gained the most (g = 0.62 for upper-body and 0.65 for lower-body strength), but adults aged 60 and over also gained significantly (g = 0.52 for both). Only the middle-aged group (45–59) showed no benefit (g = -0.10), and it contained just 108 participants. A meta-regression across all studies also found smaller strength effects at older ages. Intervention variables—daily dose above or below 5 g, loading phases, study duration, and training frequency—did not significantly change the results, and no factor consistently changed the body-composition outcomes.

Analyzing the Study: Strengths & Limitations

Strengths and Reputability

The review's protocol was prospectively registered in PROSPERO (CRD420251078957) and reported according to the PRISMA 2020 guidelines, so its planned methods were on public record before the analysis began. Pooling 63 trials and nearly 1,500 participants gives it far more precision than any single trial in this area, and the three-level model appropriately handles trials that contributed more than one result. Two reviewers independently rated each trial's risk of bias (41 of the 63 were judged low risk, 16 moderate and 6 high), and the authors tested for publication bias and reported what they found rather than leaving it out.

Limitations

1. Incremental Findings: This study primarily restates highly established findings rather than breaking new ground.
2. Low-to-Moderate Certainty: The authors' own GRADE assessment rated the certainty of evidence as low to moderate across all outcomes, so the pooled estimates are less secure than the large participant count suggests.
3. Publication Bias: The headline strength effects are likely inflated; the bias-adjusted estimates are about a third smaller.
4. Small, Short Trials: Individual trials enrolled only 8 to 45 people and lasted at most 12 weeks, so the review says nothing about longer-term use.
5. Mostly Men: Only 248 of the 1,489 participants were reported as women, against 970 men. For upper-body strength, the women-only subgroup was small and not statistically significant on its own (g = 0.45), although the difference between men and women was not significant either and the effect still favored creatine.
6. Exploratory Age Findings: The age pattern rests on unevenly represented age groups, particularly the small middle-aged subgroup, and emerged from one of many moderator analyses.
7. Uncontrolled Diet and Training: The underlying trials reported nutritional intake and training backgrounds inconsistently and incompletely.

Conflicts of Interest

No potential conflict of interest was reported by the author(s). The full text names a single funding source: a grant from the authors' own institution, Capital University of Physical Education and Sports in Beijing (Science and Technology Strengthening Project, Grant No. 155225002/011).

References

1. Wang H, Wang H, Xin X, et al. Resistance training combined with creatine supplementation: a three-level meta-analysis of multidimensional outcomes from strength enhancement to body composition remodeling. Journal of the International Society of Sports Nutrition (2026). DOI: 10.1080/15502783.2026.2718316

Glossary

1. Three-level random-effects model: A statistical technique utilized in meta-analyses to account for dependency among multiple effect sizes extracted from the same study cohort, preventing individual studies with multiple outcomes from artificially skewing the overall average. ↩

2. Meta-regression: An analytical approach that explores how specific study-level variables (like the average age of participants or the dosage provided) influence the overall magnitude of the treatment effect across diverse studies. ↩

3. PROSPERO: An international registry where researchers record a systematic review's plan before carrying it out, so readers can check that the methods were not changed after the results were known. ↩

4. GRADE: A standard system for rating how much confidence a body of evidence deserves, from "high" to "very low," based on risk of bias, consistency, directness, precision, and publication bias. ↩

5. Publication bias: The tendency for studies with positive or striking results to be published more often than those with null results, which can inflate the average effect a meta-analysis finds. ↩