Creatine in 2026: what the newest trials change about dosing, brain uptake, and safety

A 2026 systematic review and meta-analysis in Frontiers in Nutrition pooled 37 RCTs in healthy men aged 18–30 and did something most reviews avoid: it stratified by whether participants were actually resistance training.^[2] With resistance training, creatine produced significant gains in fat-free mass (+3.39 kg) and lean body mass (+2.70 kg). Without a training stimulus, neither outcome reached significance.

Anaerobic power behaved differently. Wingate peak power improved by ~71 W and mean power by ~40 W regardless of training context, consistent with the phosphocreatine system being loaded rather than remodelled. Countermovement jump improved by 2.87 cm, though heterogeneity was high (I² = 88.5%) and that estimate deserves caution.

The strength picture is more granular. A 2025 meta-analysis in Nutrients spanning 70 intervention arms across 49 studies found statistically significant but clinically small increments: bench/chest press +1.43 kg (95% CI 0.53–2.34; p = 0.002), squat +5.64 kg (p = 0.001), vertical jump +1.48 cm (p = 0.01), and Wingate peak power +47.81 W (p = 0.004).^[1]

Skeletal muscle saturates readily at 3–5 g/day. The blood–brain barrier does not cooperate as easily, and this has been the central obstacle in creatine neuroscience: most cognitive trials used muscle-appropriate doses and never confirmed that brain creatine actually changed.

Two 2025 trials addressed this directly, from opposite directions.

A randomised, double-blind, placebo-controlled trial in 36 perimenopausal and menopausal women (mean age 50.1 ± 5.7) tested creatine hydrochloride at 750 mg/day and 1,500 mg/day, plus a creatine HCl + ethyl ester combination at 800 mg/day, against placebo over 8 weeks.^[5] Medium-dose creatine HCl outperformed placebo on frontal brain creatine (+16.4% vs +0.9%; p < 0.01) and reaction time (6.6% vs 1.2% improvement; p < 0.01), and reduced self-reported mood swings and concentration difficulty. The HCl+ethyl ester arm improved alertness and anxiety scores.

The practical implication for creatine for women in the menopause transition is that a gram-scale HCl dose moved a brain imaging endpoint that gram-scale monohydrate historically has not — but with n = 36 across four arms, replication is required before this changes formulary preference away from monohydrate, which remains vastly better characterised and cheaper.

The most methodologically ambitious recent trial is a 16-week randomised, double-blind, placebo-controlled study in Experimental Gerontology (May 2026): 103 community-dwelling older adults (mean age 68.2 ± 4.6) across six arms — elastic-band power training, aquatic resistance training, or no training, each with 3 g/day creatine or placebo.^[6]

All four training groups raised serum BDNF by 8–14% while both control groups declined 4–6%. F2-isoprostanes fell 23–52% and glutathione peroxidase rose 7–20% with training; controls moved in the wrong direction on both. IL-6 fell 14–33% and TNF-α 5–20%. Creatine amplified the BDNF, antioxidant, and anti-inflammatory responses when layered on training, and creatine alone lowered F2-isoprostanes even without exercise.

Where creatine did not add value was cognition: all training groups improved on the Trail Making Test, but creatine conferred no incremental benefit over training alone. It did improve Timed Up-and-Go beyond aquatic training alone, suggesting agility and balance as a plausible target.

Two caveats the authors flag: unequal group sizes (n = 13–24) from pre-intervention attrition reduced statistical power, and the 3 g/day dose sits below the 5 g/day most protocols use — the effects may be an underestimate.

A 2026 systematic review and meta-analysis of 19 RCTs plus one double-blind crossover trial quantified the effect on renal endpoints.^[3] Creatine supplementation raised serum creatinine by a mean difference of 0.13 mg/dL (95% CI 0.07–0.18). Urea and eGFR showed no statistically significant change.

Mechanistically this is expected and benign: supplemental creatine increases the substrate pool for non-enzymatic conversion to creatinine, so the marker rises without any change in filtration. The clinically important consequence is diagnostic, not renal — a patient on creatine can present with a creatinine-based eGFR that looks impaired when it is not. Where renal function genuinely needs assessment in a supplementing patient, cystatin C–based estimation avoids the artefact entirely.

The authors are appropriately conservative: long-term RCTs beyond one year remain scarce, and patients with established chronic kidney disease should be individually assessed rather than reassured by data drawn from healthy cohorts.