People Molecules Peptides Biomarkers Studies Stem Cells Editorial Methodology
Guide

Creatine dose by goal: 5g isn't enough for the brain

5 g/day builds muscle. It does almost nothing for the brain in most people. The dose depends completely on what you are taking it for.

6 goal-specific doses5g works for muscle, not for brain20g/day x 4 weeks needed for brain uptake
Summary How much creatine should you take for muscle, brain, or sleep loss? Read the full summary ↓

For muscle, 3 to 5 grams daily is enough, with an optional 20 to 25 gram daily loading phase for 5 to 7 days. Brain uptake, meaning creatine entering the brain, may require 20 grams daily for four weeks; this raised brain creatine by 8.7%, but did not improve cognition in healthy young adults. Five grams daily may help vegetarians and older adults, whose starting levels are often lower, although the studies are limited. For acute sleep deprivation, one dose of 0.2 to 0.35 grams per kilogram, or about 14 to 25 grams at 70 kilograms, may improve mental performance after four hours, but it is not meant for daily use. Creatine has also shown benefit as an add-on to antidepressants at 5 grams daily for eight weeks. Overall, the best dose depends on the goal, and more is not automatically better.

What this means for you: For muscle, use 3 to 5 grams of creatine monohydrate daily. Higher doses are goal-specific, and brain or sleep benefits remain limited or mixed rather than guaranteed.

conflicting evidence
The brain uptake curve

Why 5 g does nothing for cognition in most people

Three measured points, one axis. Dose and duration both have to move before brain creatine does.

Swipe the chart sideways

Total brain creatine change, % from baseline+0%+4%+8%+12%5 g10 g15 g20 g25 g30 gDaily dose, grams of creatine monohydrate5 g/day x 4 weeks0%, null20 g/day x 4 weeks+8.7%, p=0.000825 g/day x 7 days-0.7% to +3.9%, nullSingle 0.2 to 0.35 g/kg bolus, taken once14 to 25 g at 70 kg, peak effect at 4 h
Read the slope, not the dose. 20 g/day for 4 weeks raised total brain creatine 8.7%, p=0.0008, with thalamus up 14.6%. Doses around 5 g/day for short durations produced no measurable change, and roughly 25 g/day for only 7 days moved brain phosphocreatine by minus 0.7% to plus 3.9%[5]. The dashed marker is a different protocol: a single 0.2 to 0.35 g/kg bolus that rescues cognition under sleep loss at 4 hours without any measurement of brain creatine[6][7].

Dose by goal

Six goals, six doses, spanning a factor of seven. Tier A means consistent randomized evidence. Tier B means one or two small trials. Tier C means the mechanism moves but the outcome has not.

Muscle and strength

3 to 5 g/day

Form: monohydrate

Duration: daily, long term

Consistent lean mass and high-intensity performance gains. Loading is optional.

Evidence tier AISSN position stand

Cognition, young healthy omnivore

20 g/day x 4 weeks

Then: 5 g/day to maintain

Brain uptake: +8.7%, p=0.0008

Raises brain creatine. Cognitive gain in this group is still not demonstrated.

Evidence tier C for cognitionSpectroscopy review

Cognition, vegetarian

5 g/day

Duration: 6 weeks, then daily

Trial: N=45 crossover, p<0.0001

Works because dietary baseline is lower. Backward digit span and Raven's matrices improved.

Evidence tier BRead the trial

Cognition, age 66 and over

5 g/day

Pooled doses: 2.2 to 20 g/day

Effect: memory SMD 0.88 vs 0.03 in the young

The one age band where a 2026 re-analysis left the pooled effect standing.

Evidence tier BRead the meta-analysis

Sleep deprivation, acute

Single 0.2 to 0.35 g/kg

At 70 kg: 14 to 25 g, once

Timing: peak at 4 h, not daily

Reversed word memory, digit span and processing speed deficits after 21 h awake.

Evidence tier BRead the trial

Depression, adjunct

5 g/day x 8 weeks

With: an antidepressant, not instead of one

Trial: N=52 RCT

Greater HAM-D improvement from week 2 onward, maintained at weeks 4 and 8.

Evidence tier BRead the trial
GoalDoseDurationEvidence qualityEffect sizeBest trial
Muscle and strength3 to 5 g/day, or 20 to 25 g/day for 5 to 7 days as a loadOngoingHighConsistent lean mass and high-intensity performance gainsISSN 2021 position stand[1][18]
Cognition, healthy young omnivore10 or 20 g/day6 weeksLow, and negativeNull: processing speed p=0.17, memory p=0.95, executive p=0.59Moriarty 2023, N=30[2]
Cognition, vegetarian5 g/day6 weeksModerateWorking memory and Raven's matrices, p<0.0001Rae 2003, N=45[3]
Cognition, older adult 66 to 765 to 20 g/day5 days to 24 weeksModerateMemory SMD 0.88 versus 0.03 in the youngProkopidis 2023 meta-analysis[4]
Acute sleep deprivationSingle 0.2 to 0.35 g/kg bolus, 14 to 25 g at 70 kgOne dose, peak at 4 hModerateReversed word memory, digit span and processing speed deficitsGordji-Nejad 2024, N=15[6]; replication N=29[7]
Depression, SSRI adjunct5 g/day8 weeksModerateGreater HAM-D improvement from week 2Lyoo 2012, N=52[8]

All doses are creatine monohydrate. Split anything above 5 g into separate servings.

The mechanism: muscle saturates, brain does not

Myth buster

"5 g a day covers everything"

  1. 5 g fills muscle in 3 to 5 days. Maintenance dosing of 3 to 5 g/day, or 0.1 g/kg/day, is what saturates skeletal muscle. That is where the number comes from[1].
  2. Brain uptake is slower. Brain creatine typically rises only 5 to 15% even under sustained supplementation[12].
  3. Transport across the blood-brain barrier is the limit, not how much powder is in the glass. Higher doses for longer periods are needed to move brain concentrations at all[12].
  4. The working protocol is 4 times the dose for 4 times the duration: 20 g/day for 4 weeks instead of 5 g/day for 5 days. That combination gave +8.7%, p=0.0008. A single 20 g dose gave 3 to 7.7% and missed significance[5].
  5. Vegetarians and older adults respond at 5 g because their starting point is lower. 45 vegetarians improved working memory at p<0.0001[3], and pooled memory effect was SMD 0.88 in adults aged 66 to 76 against SMD 0.03 in those aged 11 to 31[4].

The 3 to 5 g figure comes from muscle physiology. The ISSN describes maintenance as 3 to 5 g/day or 0.1 g/kg/day, with an optional 20 to 25 g/day load for 5 to 7 days[1]. That number saturates skeletal muscle. It was never derived from a brain endpoint.

Brain creatine behaves differently. Muscle sits behind a transporter that fills quickly at gram doses. Brain uptake is slower and the tissue is already tightly regulated, so brain creatine typically rises only 5 to 15% even under sustained supplementation[12]. Reviews of the magnetic resonance spectroscopy data report that lower doses around 5 g/day for short durations produce no noticeable change in brain creatine[5].

The cognitive trials match that. Moriarty ran 10 g/day against 20 g/day against placebo for 6 weeks in 30 healthy young adults and found nothing: processing speed p=0.17, episodic memory p=0.95, executive function p=0.59[2]. A 2018 systematic review put it plainly, that performance on cognitive tasks stayed unchanged in young individuals[13].

Two groups do respond at 5 g/day, and both were starting lower. Vegetarians eat almost no dietary creatine, and 45 of them on 5 g/day for 6 weeks improved backward digit span and Raven's matrices at p<0.0001[3]. A preregistered replication of the same protocol ran at N=123[15]. Older adults are the second group: pooled memory effect was SMD 0.88 in adults aged 66 to 76 against SMD 0.03 in those aged 11 to 31[4], and the single largest contributing trial gave 20 g/day for 7 days to 32 adults averaging 76 years[14]. A 2026 commentary re-analyzed that meta-analysis after correcting a unit-of-analysis error and found older adults were the only subgroup where the effect survived[16]. That is worth knowing before treating the pooled numbers as settled.

The 20 g/day protocol for brain uptake

If the goal is measurably more creatine in brain tissue, the dose is about 20 g/day and the duration is weeks, not days. Spectroscopy measurements summarized by Fabiano and Candow give the shape of it: 20 g/day for 4 weeks raised total brain creatine 8.7%, p=0.0008, with thalamus up 14.6% and white matter up 11.5%. A single 20 g dose moved it 3 to 7.7% and missed significance[5].

Loading alone is not enough. Solis gave roughly 25 g/day for 7 days and measured a change of minus 0.7% to plus 3.9% in brain phosphocreatine, which is nothing[5]. Seven days is a muscle protocol. The brain needs the four weeks.

Split the 20 g into four servings of 5 g. In a 28-day trial of 59 soccer players, a single 10 g dose produced diarrhea in 55.6% of participants against 28.6% on split 5 g doses, p<0.05, and split dosing did not differ from placebo[10]. Expect 1 to 3 kg of body weight in the first week at loading doses, mostly total body water[10].

One honest caveat: raising brain creatine and improving cognition are two different claims. The 20 g/day protocol has strong support for the first and weak support for the second outside acute stress and older age[2][17].

Acute protocol for sleep loss

This is the one place where a very large single dose has direct trial support. Fifteen adults kept awake for 21 hours took a single 0.35 g/kg dose, roughly 24.5 g for a 70 kg adult. It reversed the sleep-deprivation deficits in word memory, digit span and processing speed, with the maximum effect at 4 hours and measurable effects out to 9 hours[6]. A 29-person replication at 0.2 g/kg, mean 14.1 g, reproduced the result at the lower dose[7].

Three rules follow. First, this is a single bolus taken during the sleep-deprived window, not a daily regimen. Second, time it about 4 hours ahead of the work that matters. Third, it is 3 to 7 times the muscle dose, so the gastrointestinal cost is real and the water-weight shift is immediate[10].

The older approach was 20 g/day for 7 days before a 24-hour deprivation protocol, which also reduced decrements in reaction time, balance and mood[12]. The single-dose route is simpler and now has two trials behind it.

Contraindications and non-indications

PopulationDose that failedTrial
Parkinson's disease10 g/day for a median 4 yearsN=1,741, 45 sites, stopped early for futility, JAMA 2015[9]
Cognition in healthy, well-rested young adults10 g/day and 20 g/day for 6 weeksN=30 dose-response RCT, null on all three domains: processing speed p=0.17, memory p=0.95, executive p=0.59[2]
Anyone taking one large servingSingle 10 g doseDiarrhea in 55.6% against 28.6% on 2 x 5 g, p<0.05, in 59 soccer players over 28 days[10]

What the safety data show

  • 1 to 80 g/day for 5 days to 60 months produced no kidney impairment on estimated GFR, serum and urinary creatinine, urea, proteinuria or albuminuria[10]. The cohorts were mostly healthy young men.
  • No hair loss signal. The claim rests on one 2009 study of 20 rugby players and has not been replicated since[11][1].
  • Water weight is 1 to 2 kg of intramuscular and total body water in the first week at loading doses, not fat[10].
  • Serum creatinine rises without kidney damage. Ask for cystatin C or a direct GFR measure if you already have reduced kidney function[10].

Parkinson's disease: no. The NIH NET-PD LS-1 trial randomized 1,741 patients with early treated Parkinson's to 10 g/day or placebo across 45 sites, with a median 4 years of follow-up. It was stopped early for futility[9]. This is the largest creatine trial ever run and it is negative.

Depression: adjunct only. 5 g/day for 8 weeks alongside escitalopram in 52 women produced greater HAM-D improvement from week 2 onward[8]. Creatine was added to treatment, not substituted for it.

Hair loss: unsupported. The claim comes from one 2009 study of 20 rugby players in which dihydrotestosterone rose 56% after loading[11]. It has not been replicated, the values stayed within normal clinical limits, baseline DHT was 23% lower in the creatine group, and no study has reported hair loss in humans[1].

Kidneys: safe in healthy adults. Kidney-function studies span 1 to 80 g/day for 5 days to 60 months and consistently found no impairment on estimated GFR, serum and urinary creatinine, urea, proteinuria or albuminuria[10]. The cited cohorts were mostly healthy young men. If you already have reduced kidney function, ask for cystatin C or a direct GFR measure, because creatine supplementation raises serum creatinine without any change in kidney function[10].

How much to spend

Creatine monohydrate is the only form with the evidence base. Premium forms sold at a markup, hydrochloride, ethyl ester and buffered creatine, have no demonstrated advantage over monohydrate on muscle or cognitive endpoints[1]. Every trial cited on this page used monohydrate.

That makes the buying decision short: pick a tested by independent lab monohydrate and pay for grams, not for a proprietary name. A 20 g/day brain protocol burns four times the powder of a muscle dose, so price per gram is the number to compare.

Creatine Monohydrate
Thorne HealthTech
5 g creatine per serving
Muscle dose met in one serving
Score 8.9/10
Creatine Monohydrate Powder
DoNotAge
10 g creatine per serving
Muscle dose met in one serving
Score 8.6/10
AG1 Pro
AG1 (Athletic Greens)
5 g creatine per serving
Muscle dose met in one serving
Score 7.3/10
Daily Ultimate Essentials
IM8
3 g creatine per serving
Below 5 g, top up for the muscle dose
Score 6.9/10
Longevity Mix - Blood Orange
Blueprint
2.5 g creatine per serving
Below 5 g, top up for the muscle dose
Score 6.7/10
Creatine+
Elysium Health
5 g creatine per serving
Muscle dose met in one serving
Score 6.5/10

Ranked by our six-pillar overall score. See the full ranking: Best Creatine Supplements 2026.

Frequently asked questions

Do I need to take creatine with a loading dose?

Not for muscle. Loading at 20 to 25 g/day for 5 to 7 days saturates muscle faster, and 3 to 5 g/day gets to the same place over a few weeks. Loading does not reliably move brain creatine: 25 g/day for 7 days changed brain phosphocreatine by minus 0.7% to plus 3.9%.

Is 5 g of creatine enough for the brain?

For most people, no. 5 g/day for short periods produced no measurable change in brain creatine. It worked in vegetarians, whose baseline is lower, at p<0.0001 over 6 weeks, and it worked as a depression adjunct. In healthy young omnivores, 10 and 20 g/day for 6 weeks were both null on cognition.

Does creatine cause hair loss?

No human study has reported hair loss from creatine. The claim traces to one 2009 study of 20 rugby players that measured a rise in dihydrotestosterone. It has not been replicated in the years since, and the values stayed within normal clinical limits.

Is creatine safe for kidneys?

Kidney studies span 1 to 80 g/day for 5 days to 60 months and found no impairment on estimated GFR, serum and urinary creatinine, urea, proteinuria or albuminuria. Anyone with existing kidney disease should be assessed with cystatin C or direct GFR, because creatine shifts serum creatinine independently of kidney function.

How much creatine for cognition?

It depends on who you are. Raising brain creatine takes about 20 g/day for at least 4 weeks. Acute sleep-loss rescue takes a single 0.2 to 0.35 g/kg dose. Vegetarians and adults aged 66 to 76 respond at 5 g/day. Well-rested young omnivores have not been shown to respond at any dose tested.

Related pages

Corrections and source queries: [email protected].

References

  1. [1] Antonio et al. 2021. ISSN position stand, common questions and misconceptions about creatine. J Int Soc Sports Nutr 18:13. PMID 33557850. https://pmc.ncbi.nlm.nih.gov/articles/PMC7871530/
  2. [2] Moriarty et al. 2023. Creatine 10 g vs 20 g vs placebo for 6 weeks in healthy young adults. Brain Sciences 13:1276. PMID 37759877. https://pmc.ncbi.nlm.nih.gov/articles/PMC10526554/
  3. [3] Rae et al. 2003. Creatine 5 g/day for 6 weeks in vegetarians. Proc Biol Sci. PMID 14561278. https://pubmed.ncbi.nlm.nih.gov/14561278/
  4. [4] Prokopidis et al. 2023. Creatine and memory, meta-analysis of 10 RCTs. Nutrition Reviews 81:416-427. PMID 35984306. https://pubmed.ncbi.nlm.nih.gov/35984306/
  5. [5] Fabiano and Candow 2025. Creatine and brain bioenergetics review. J Psychiatry Brain Sci 10(4):e250006. https://jpbs.hapres.com/htmls/JPBS_1766_Detail.html
  6. [6] Gordji-Nejad et al. 2024. Single 0.35 g/kg creatine dose under 21 h sleep deprivation, N=15. Scientific Reports. https://www.nature.com/articles/s41598-024-54249-9
  7. [7] Gordji-Nejad et al. 2026. Single 0.2 g/kg creatine dose replication, N=29. Nutrients 18:1192. PMID 42075005. https://pmc.ncbi.nlm.nih.gov/articles/PMC13119191/
  8. [8] Lyoo et al. 2012. Creatine 5 g/day as escitalopram augmentation in major depression, N=52. Am J Psychiatry. PMID 22864465. https://pubmed.ncbi.nlm.nih.gov/22864465/
  9. [9] Kieburtz et al. 2015. NET-PD LS-1, creatine 10 g/day in 1,741 Parkinson's patients, stopped for futility. JAMA 313:584-593. https://jamanetwork.com/journals/jama/fullarticle/2108890
  10. [10] Frontiers in Nutrition 2025. Creatine monohydrate safety review, kidney and gastrointestinal data. https://www.frontiersin.org/journals/nutrition/articles/10.3389/fnut.2025.1682746/full
  11. [11] van der Merwe et al. 2009. Creatine and DHT in 20 rugby players. Clin J Sport Med. PMID 19741313. https://pubmed.ncbi.nlm.nih.gov/19741313/
  12. [12] Kreider et al. 2017. ISSN position stand, creatine supplementation in exercise, sport and medicine. https://pmc.ncbi.nlm.nih.gov/articles/PMC5469049/
  13. [13] Avgerinos et al. 2018. Creatine and cognition, systematic review of 6 RCTs. Exp Gerontol. PMID 29704637. https://pubmed.ncbi.nlm.nih.gov/29704637/
  14. [14] McMorris et al., creatine 20 g/day for 7 days in adults aged 76, summarized in the ISSN evidence base. https://pmc.ncbi.nlm.nih.gov/articles/PMC6093191/
  15. [15] Sandkuhler et al. 2023. Preregistered replication of the 5 g x 6 weeks protocol, N=123. BMC Medicine 21:440. https://pmc.ncbi.nlm.nih.gov/articles/PMC10647179/
  16. [16] Citherlet 2026. Commentary on unit-of-analysis errors in creatine cognition meta-analyses. Frontiers in Nutrition. https://www.frontiersin.org/journals/nutrition/articles/10.3389/fnut.2026.1716285/full
  17. [17] Turner et al. 2015 and related acute-stressor trials, summarized in the creatine and brain function review. https://pmc.ncbi.nlm.nih.gov/articles/PMC10721691/
  18. [18] Kreider et al. 2021. ISSN position stand summary on creatine and exercise performance. PMID 34234088. https://pubmed.ncbi.nlm.nih.gov/34234088/