Supplements · Metabolic & Cardiometabolic
creatine increases muscle strength and power
In plain terms: Does creatine actually make you stronger and more powerful?
Strong support Supplements 💰 Industry COI noted
Part of: 🧪 creatine
RefutedContestedStrong support
consensus score 0.92
Yes — this is one of the best-established supplement effects there is. Paired with resistance training, creatine reliably adds strength and power across ages and both upper and lower body. The boost is modest but real and well-replicated.
📅 Last reviewed: 2026-07-14 ⓘ
Evidence ladder
How far up the ladder this claim has climbed. A high consensus on a low rung means "consistent so far," not "proven in people."
Top evidence so far: All trials, pooled (Meta-analysis)
MechanismIn-vitroAnimalObservationalRCTMeta-analysis
How the studies fall
12 support 0 contradict 0 tested null 1 mixed · 13 sources, 12 independent groups
What the evidence shows
Robust, consistent meta-analytic support for creatine + resistance training on maximal strength and power; effect sizes are modest but reliable across ages and both upper and lower body.
The evidence (14)
| Source | Grade | Stance | Quality | Finding |
|---|---|---|---|---|
| Fernandez-Garrido J et al 2026 · Experimental Gerontology | RCT | supports | low | 16-wk randomized, double-blind, placebo-controlled trial, 103 community-dwelling older adults (68.2 +/- 4.6 y) allocated across six arms - elastic-band or aquatic high-load velocity-intentional training or no-training control, each crossed with creatine 3 g/d or placebo (NCT06620666), so ~17 per cell. Creatine was the randomised variable. 'Creatine provided additional benefits, increasing GPx, reducing IL-6/TNF-alpha and improving strength and function when combined with exercise modalities.' The measured strength endpoints were isokinetic knee/elbow at 60 deg/s and handgrip - no POWER endpoint is among the reported outcomes, so this source carries the strength half of the claim only. No effect size or p-value is given for the creatine contrast; the reported d = 0.12-1.18 belongs to the aquatic-vs-elastic training comparison, not to creatine. Creatine and publication charges were supplied by manufacturer AlzChem Group AG, with two authors in a consulting relationship with it. |
| Lanhers 2017 · Sports Med | meta-analysis | supports | moderate | Systematic review + meta-analysis of 53 randomized placebo-controlled trials (563 creatine / 575 control): 'Results did not differ at T0, while, at T1, the effect size (ES) for bench press and chest press were 0.265 (95 % CI 0.132-0.398; p < 0.001) and 0.677 (95 % CI 0.149-1.206; p = 0.012), respectively'; 'Overall, pectoral ES was 0.289 (95 % CI 0.160-0.419; p = 0.000), and global upper limb ES was 0.317 (95 % CI 0.185-0.449; p < 0.001)'. Effect is real but SMALL and scoped to UPPER-LIMB strength in exercises <3 min - no lower-body or distinct power endpoint. Quality moderate not high: the headline analysis pools post-supplementation (T1) values rather than randomised change scores ('Meta-analysis of changes between T0 and T1 gave similar results' - no numbers given), no heterogeneity, risk-of-bias or publication-bias statistic is available (abstract-only; paywalled, no OA full text), and the paper's own conclusion overreaches by reading a null meta-regression as 'demonstrating the efficacy of creatine independently of all listed conditions'. |
| Naddafha S, Antonio J, Kreider RB, Stout JR 2026 · J Int Soc Sports Nutr | meta-analysis | supports | moderate | Leg-press 1RM pooled from 3 RT trials (k=3, n=111 — not the whole 7-RCT/n=608 review): MD +7.5 kg (95% CI +2.2 to +12.8, I2=0%), ~6-8% above training alone; authors' GRADE moderate certainty. Benefit confined to trials combining creatine >=5 g/day with resistance training; the two trials without exercise showed no strength difference. Upper-body (bench-press 1RM) was null: pooled MD ~+3.1 kg (95% CI -0.5 to +6.7). |
| Wang Y et al 2026 · Scientific Reports | RCT | supports | low | 4-arm randomised, double-blind, placebo-controlled trial; n=60 physically active men (~15/arm, age 21+/-2.8 y), 4-day loading: creatine alone (CR, 0.3 g/kg/d), creatine+carbohydrate (CRCHO), creatine+carbohydrate+whey protein (CRCPS), or xylitol placebo (PLA). Creatine ITSELF was the randomised variable (PLA received no bioactive ingredient), so the design does bear on a creatine claim. Significant group x time interactions for mean and peak power at all three 30-s Wingate sprints (partial eta-sq 0.16-0.72). Within-group vs baseline: CR raised absolute mean power in AWT1 (p<0.01), AWT2 and AWT3 (p<0.05) and relative mean power in AWT1 and AWT3 only; all three supplement arms raised absolute and relative peak power at all three sprints (p<0.01); PLA fell in absolute mean power at AWT2 and AWT3 (p<0.01). ENDPOINT CAVEAT: the placebo decrement is in MEAN power only - PLA itself significantly INCREASED absolute and relative peak power at AWT3 (p<0.01), so the peak-power half of this claim is not separated from a learning/familiarisation effect. Inference is within-group pre-post: the authors state 'direct statistical comparisons between supplementation groups did not reveal significance' and name modest sample size, absent strict dietary monitoring and short duration as limitations. No funding; no competing interests declared; no trial-registry identifier appears in the text read. |
| Ben Maaoui K et al 2026 · Nutrients | RCT | supports | low | Acute single-dose timing trial, NO loading phase: 11 physically active men, randomised placebo-controlled 5-condition crossover (0.1 g/kg creatine before / during / after exercise, maltodextrin placebo, no-supplement control; one session per condition, 7-day washout). Endpoint is total external load lifted across six sets at 80% 1RM in bench press and back squat within a single session - not 1RM and not chronic strength gain. Creatine-before beat placebo (bench press p=0.006, d=0.56; back squat p<0.001, d=0.49) and beat the no-supplement control (p=0.014, d=0.28; p<0.001, d=0.24); the p=0.035 / p<0.001 figures are the omnibus condition F-tests, not these contrasts. The title's 'more than during or after' rests on ONE significant timing contrast (back squat before-vs-during, d=0.20, p=0.026) - before-vs-after was null in both lifts and no timing contrast was significant in the bench press. Placebo was the worst arm overall: the no-supplement control also out-performed it in both lifts, so the margin is not cleanly attributable to creatine. No effect on countermovement jump power, height, force or speed (all p>=0.29). Pilot; blinding held only for the pre-exercise arm. |
| Deng B, Li H, Chen C, He J, Zhang W, Lin H 2025 · Food Science & Nutrition 2025;13(12):e71243 | meta-analysis | supports | low | Network meta-analysis of 30 RCTs in trained athletes (693 athletes across six supplement nodes; PROSPERO CRD420251106522). The creatine estimates come from the per-outcome networks, not the full 693: peak power 'Seventeen studies encompassing 427 athletes'; mean power 'Fourteen studies involving 293 athletes'. Creatine beat placebo on 30-s Wingate peak anaerobic power - 'Moderate-certainty evidence likewise indicated significant gains with creatine (SMD = 0.62, 95% CI 0.20-1.03; SUCRA = 62.5%)' - and on mean anaerobic power at 'Very-low-certainty evidence ... creatine: SMD = 0.74, 95% CI 0.08-1.39; SUCRA = 71.0%'. No effect on VO2max (SMD 0.11, 95% CI -0.43 to 0.64) and none on endurance performance, where creatine ranked last of six (SMD -0.46, 95% CI -1.25 to 0.32; SUCRA 10.7%). Endpoint is Wingate power output, not maximal strength - no 1RM outcome is in this network, so the row bears on the power half of the claim only. Quality lowered to low: 'three studies (10%) were judged to be at low risk of bias, 22 studies (80%) at unclear risk, and three studies (10%) at high risk', and Appendix S2 enters the creatine evidence as six rows from about four distinct trials ('Li et al. 2005a/b' are two identical rows of the same citation; 'Bemben et al. 2001a/b' reuse one 9-man creatine arm against two different control arms). The mean-power confidence intervals printed for creatine and protein in the Results text are transposed relative to league Table S7.2; the peak-power creatine figure matches the league table exactly. |
| Devries MC, Phillips SM 2014 · Med Sci Sports Exerc | meta-analysis | supports | moderate | Meta-analysis of randomized placebo-controlled trials, 357 older adults, 12.6 ± 4.9 wk of resistance training. The strength result is split, not uniform: 'Cr + RT increased chest press (P = 0.004) and leg press (P = 0.02) one-repetition maximum to a greater extent than RT alone, with no difference in the effect on knee extension or biceps curl one-repetition maximum, isokinetic or isometric knee extension peak torque.' Two of six strength endpoints moved — the multi-joint compound lifts — and the single-joint and isokinetic measures did not. No power endpoint (jump, sprint, rate of force development) was pooled, so the claim's power arm is untested here. Authors: 'the limited number of studies indicates further work is needed.' |
| Kazeminasab F, Kerchi AB, Sharafifard F, Zarreh M, Forbes SC, Camera DM, Lanhers C, Wong A, Nordvall M, Bagheri R, Dutheil F 2025 · Nutrients | meta-analysis | supports | moderate | Meta-analysis of 69 RCTs (89 arms, 1937 adults, PROSPERO CRD420251025716). Creatine beat placebo on 4 of 6 pooled endpoints — 'bench and chest press strength [WMD = 1.43 kg, p = 0.002], squat strength [WMD = 5.64 kg, p = 0.001], vertical jump [WMD = 1.48 cm, p = 0.01], and Wingate peak power [WMD = 47.81 Watts, p = 0.004]' — but was null on the other two: 'no significant differences in handgrip strength [WMD = 4.26 kg, p = 0.10] and leg press strength [WMD = 3.129 kg, p = 0.11]'. The authors' own summary: 'benefits were not uniform across all outcomes or subgroups'. Effects concentrated in younger trained males; older adults were null on all three lifts (bench p = 0.26, leg press p = 0.71, squat p = 0.73) and women null on squat/leg press/jump/Wingate. Quality moderate, not high: the paper's own PEDro audit records that '63 studies failed to score on both' participant blinding and intention-to-treat, Egger's test showed significant publication bias for bench press (p = 0.0003) and leg press (p = 0.03), and co-author SCF discloses advisory roles with a creatine manufacturer. |
| López-Moreno M et al 2026 · European Journal of Applied Physiology | RCT | mixed | moderate | 20 physically active vegan adults (10 per arm), 4wk creatine monohydrate 0.1 g/kg/day vs placebo: 'Performance improvements were observed in lower-body explosive performance, particularly in CMJ, as well as in maximal height and flight time during repeated jumps (p < 0.001-0.04), with no changes in sprint or strength tests' - the null covering handgrip and isometric mid-thigh pull. Split result, hence mixed. Note the authors' own conclusion overstates this ('enhances muscle strength and lower-body muscular power'); the results sentence governs. Nulls at 10 per arm are underpowered, not demonstrated absence. |
| Wang Z et al 2026 · Nutrients | meta-analysis | supports | moderate | Network meta-analysis of 35 RCTs (1211 trained athletes, >=6 months structured training); the muscle-strength network used 28 of those trials, with creatine vs placebo informed by 11 trials (creatine arms n=386, placebo n=496). Creatine vs PLACEBO: SMD = 0.46 (95% CI 0.29 to 0.63), SUCRA 82.4%; robust to dropping high-risk-of-bias trials (SMD = 0.42, 95% CI 0.23 to 0.61). Authors' GRADE for this comparison is MODERATE, downgraded for suspected publication bias (funnel asymmetry, Egger p = 0.08); review not PROSPERO-registered. Creatine beat omega-3 (SMD 0.30) but NOT protein (SMD 0.13, 95% CI -0.09 to 0.35, p = 0.25), and 'There are no head-to-head trials evaluating the relative efficacy of protein, creatine, and omega-3 supplementation' - all active-vs-active estimates are indirect and GRADE low. Pooled 'strength' = 1RM, isokinetic peak torque, maximal voluntary contraction; no separate pooled power outcome is reported. |
| Branch 2003 · Int J Sport Nutr Exerc Metab | meta-analysis | supports | moderate | Meta-analysis of 100 randomized, placebo-controlled, subject-blinded studies. Effects on high-intensity/short-duration performance are SMALL, not small-moderate: 'ATP-PCr (n=17, 0.24 +/- 0.02), G (n=135, 0.19 +/- 0.05)', with laboratory resistance-type tasks at 0.25 +/- 0.02 versus field tasks 0.14 +/- 0.04 (p=.014) and upper body 0.42 +/- 0.07 versus lower 0.21 +/- 0.02. The paper reports no separate strength endpoint — strength is read off the conclusion that 'ES was greater for changes in lean body mass following short-term CS, repetitive-bout laboratory-based exercise tasks < or = 30 s (e.g., isometric, isokinetic, and isotonic resistance exercise)'. The same analysis finds 'CS does not appear to be effective in improving running and swimming performance', so the benefit is specific to short laboratory strength/power tasks. |
| dos Santos MSC et al 2026 · Front Nutr | RCT | supports | low | Randomised single-blind placebo-controlled pilot in post-COVID-19 condition (58 analysed; creatine 6 g/day vs 18 g/day vs maltodextrin placebo, exercise 3x/week in ALL arms, 4 weeks). Muscle strength was a SECONDARY outcome; the trial's a-priori primary endpoint was fatigue on the PFS-R. Handgrip strength (dominant limb) rose in the 6 g/day arm only: 'Additionally, this same group showed a significant increase in handgrip strength [mean difference (DiD estimate): 4.40 kgf; 95% CI: 0.27 to 8.52; p = 0.037].' The 18 g/day arm was flat (DiD 0.67 kgf, 95% CI −3.59 to 4.94; p = 0.751), and respiratory muscle strength (MIP, MEP) was null in both arms. Direction favours the claim, but weakly: the authors write that 'Given the exploratory nature of secondary outcomes and the number of comparisons performed, findings beyond the primary outcome should be interpreted cautiously and confirmed in adequately powered trials.' |
| Kreider RB, Kalman DS, Antonio J, Ziegenfuss TN, Wildman R, Collins R, Candow DG, Kleiner SM, Almada AL, Lopez HL 2017 · J Int Soc Sports Nutr | observational | supports | low | ISSN position stand (narrative review; Candow co-author, CRN-funded). Position statement 1: 'Creatine monohydrate is the most effective ergogenic nutritional supplement currently available to athletes with the intent of increasing high-intensity exercise capacity and lean body mass during training.' Body text carries the strength/power reading - creatine lets an athlete do more work per set or sprint, 'leading to greater gains in strength, muscle mass, and/or performance due to an improvement in the quality of training' - and Table 1 lists 'Increased muscle mass & strength adaptations during training'. The quantified figure is performance, not strength: 'After creatine loading, performance of high intensity and/or repetitive exercise is generally increased by 10-20% depending on the magnitude of increase in muscle PCr'. No pooled effect size and no data of its own; a summary of others' trials. |
| Dos Santos 2021 · Nutrients | meta-analysis | supports | low | Systematic review + random-effects meta-analysis of 8 RCTs, n = 176 older FEMALES (mean age 56-70), creatine + RT vs placebo + RT, 12-52 weeks (PROSPERO CRD42020221648). Positive but partial: 'Overall, Cr resulted in a significant increase in upper-body strength ( p = 0.04; I 2 = 0%; SMD = 0.35 [95% CI 0.02-0.69]).' Lower body was null overall: 'Overall, there was no significant effect from Cr on measures of lower-body strength ( p = 0.18; I 2 = 0%; SMD = 0.22 [95% CI -0.10-0.55]).' Both regions turned significant only in trials >=24 weeks (upper SMD 0.41, p = 0.05, 4 studies n = 97; lower SMD 0.44, p = 0.03, 4 studies n = 100). Only 1RM chest/bench press and leg press/hack squat were pooled - no power outcome was measured. Quality downgraded to low on the authors' own certainty rating: 'we judged the overall certainty of the evidence in this review as low, after downgrading the scores due to issues related to imprecision (i.e., total cumulative sample size lower than 300).' Median PEDro 7, but four trials had >15% withdrawal and six did not report intention-to-treat. |
Disagree, or know a study we missed?
We grade by evidence, not opinions. The way to weigh in is to point us to a study we haven't cited (check the evidence table above first), or to flag a problem with one we have. Every submission is reviewed; if it holds up, the grade updates and shows in Science Changes Its Mind.
Opens a short form. You'll sign in with Google so submissions are tied to a real account — we don't display your identity, and we only accept a link we can verify (PubMed, DOI, ClinicalTrials.gov).
Educational only, not medical advice. Grades and scores reflect published evidence weighted by study design and quality; see the methodology.