This chapter assembles what the corpus says about creatine monohydrate: the phosphagen mechanism and the competing account of cell swelling as a direct growth signal, the case for a flat 5 g/day with loading as optional, why every "improved" form has failed to beat the commodity powder, and what the pooled safety data does and does not cover. It then works through the two places the evidence is genuinely contested — the brain-dosing threshold and a 2026 null trial read two opposite ways — and closes on effect size, which is real but smaller than the lifting internet believes.
Creatine works by supplying phosphocreatine, which acts as a phosphate donor to regenerate ATP fast. That maps onto one energy system in particular — the phosphagen system, the roughly ten-second all-out effort, a 40-yard dash — and not onto the oxidative/aerobic pathway. The performance data tracks the mechanism cleanly: Creatine Supplementation improves power, hypertrophy and high-intensity anaerobic output by roughly 5–10%, sometimes up to 15%, while a meta-analysis found no endurance benefit in trained athletes. Creatine does not augment aerobic capacity, and no source in this corpus claims otherwise.
Where the corpus does split is on how that ATP buffer becomes muscle. One account is indirect and training-mediated: creatine adds one to two extra reps per set rather than directly boosting muscle protein synthesis in sedentary people, so the strength and size gains are really the downstream product of the extra training volume it enables. On that reading creatine is a lever on the training stimulus described in Resistance Training & Performance Supplements, not an anabolic agent in its own right.
The other account is direct. Creatine is an osmolyte: it pulls water into the muscle fiber, raising total body water and intracellular water specifically, not extracellular water. Per RP Strength's Dr. Mike, that intracellular cell swelling is itself a growth signal, sensed through the SWELL1/VRAC → mTORC1 pathway, and it operates independently of taking on more volume, training closer to failure, or doing more reps — he describes creatine's growth effect as roughly nine-out-of-ten reliable, against the unreliable direct effect of pump supplements, which mostly cause vasodilation of the vessels and interstitial space around the fiber rather than true swelling inside it. Layne Norton frames the same point from the other side: muscle cells are about 70% water, so more intracellular fluid means bigger cells by default, which makes water retention inseparable from the anabolic mechanism rather than an unwanted side effect of it.
One group gets a disproportionately large response: vegetarians and vegans, who have no dietary meat source and rely entirely on endogenous synthesis, so supplementation moves them further from a lower baseline. And one thing is missing across all of it — there is no biomarker for individual tissue saturation, so dosing stays protocol-based rather than personally titrated.
The classic protocol — 20–30 g/day for a week or two, inherited from 1980s gym culture and the belief that "creatine phosphate" beat monohydrate — is outdated. A flat 5 g/day with no loading saturates muscle within 3–4 weeks with less GI distress, and effectiveness plateaus at 3–5 g/day: once stores are saturated, higher doses don't outperform it and the excess is simply excreted in urine. Loading is only genuinely useful ahead of imminent competition, when time-to-saturation matters.
The cleanest test of that is inside Creatine Supplementation: in a 2026 randomized four-arm trial in resistance-trained men, the group that ran a 5-day load at 0.3 g/kg/day before dropping to maintenance showed no outcome difference at week 8 from the monohydrate group that never loaded. Loading changes when you reach saturation, not the end-state strength or cross-sectional area. Skipping it costs about two extra weeks and nothing else.
Bryan Johnson's team frames the same target as bodyweight-scaled rather than flat — roughly 0.1 g/kg/day, about 5–7 g for an average adult — explicitly in preference to the 20 g/day loading week, which they note carries more water-retention side effects. The same 0.1 g/kg scaling is the recommendation for pediatric and adolescent dosing instead of a flat adult dose. (Their own packaged product, the Blueprint 'Longevity Mix', bundles only 2.5 g as part of a multi-supplement stack — a stack component, not a dosing protocol.) The Andy Galpin protocol recorded in Creatine Monohydrate lands in the same place at 3–5 g/day with no loading, though that page carries a warning banner: it was written from model recall rather than through this corpus's normal acquisition pipeline, so treat it as corroborating rather than as an independent anchored source.
For GI discomfort, the recommended fix is splitting the dose rather than switching forms: half in the morning and half in the evening, or two 2.5 g doses, or even 1–2 g increments, using micronized monohydrate and adequate water. Notably, in the acute sleep-deprivation study discussed later in this chapter, no participant reported GI side effects at a single ~30 g dose. Higher chronic doses have been tested too — up to 30 g/day for as long as five years without significant downsides.
Creatine monohydrate is the most tested, most efficacious and cheapest form, and every alternative in this corpus is argued to be a commercial play rather than an efficacy gain. The economics are the explanation: monohydrate has 30–40 years of evidence and is widely manufactured, which has driven its price down and stripped the margin out of it, creating a recurring incentive to invent a new alleged flaw of monohydrate and sell the fix.
The cycle has run three times so far, per Creatine Supplementation:
One marketing line gets a specific rebuttal. Products advertised as "not causing water retention" are, by Norton's logic, tacitly claiming not to be anabolic, since intracellular water is part of the mechanism — and no evidence supports the no-water-retention claim for those products anyway. His formulation: "If you are saying that your creatine does not cause water retention, you are actually saying that your creatine is not anabolic." For the mixing complaint that drives people to HCl, two cheaper fixes stay inside monohydrate: dissolve it in an acidic liquid such as orange juice, where it is more soluble and stable, or buy micronized powder.
Getting it from food is impractical rather than merely inconvenient. Red meat carries only ~1–1.5 g of creatine per pound and roughly half is destroyed by cooking, which puts the equivalent of a standard supplemental dose at about 7 lbs of red meat a day — far more expensive than the powder. That directly rebuts the claim, made by Paul Saladino on The Joe Rogan Experience, that a meat-based diet makes supplementation unnecessary.
A form debate assumes the product contains what the label says, and for creatine that assumption fails in one format specifically. Roughly 95% of tested creatine gummy products were found to contain essentially no real creatine. The best-known case, from the testing lab Light Labs, is the top-selling creatine gummy on Amazon testing at 0% of its claimed content — while carrying positive consumer reviews, which is explained as placebo: creatine's physiological effects aren't something you can verify from one day to the next by feel.
The economics make the fraud legible. Gummies retail around $40, and creatine is the most expensive component in them, so omitting it is where the margin is. Gummies also have more failure points across manufacturing and shelf life than a scoop in a bag does, and capsules and other poorly manufactured formats are described as unreliable for similar reasons. Plain powder is the low-fraud, low-degradation format.
The practical filter that falls out of Creatine Supplementation is narrow: plain monohydrate powder, NSF-certified or Creapure, with creatine monohydrate as the sole ingredient on the label aside from flavoring, and no fillers. That is the same third-party-verification logic developed at length in Evaluating Evidence & Supplement Quality, applied to the one supplement in this corpus where the underlying evidence is strongest — which is precisely why the label risk, rather than the efficacy risk, is the binding constraint here.
The headline safety evidence is a pooled meta-analysis spanning more than 13,000 participants, doses from a few grams up to 30 g/day, ages from teens to 77, and durations from a few weeks to 14 years. Adverse-event reporting rates were statistically indistinguishable between creatine and placebo — about 13% in each arm — including for the two effects people report most, cramping and GI distress. The screened list was wide: vertigo, hypertension, headache, dizziness, nausea, diarrhea, impaired concentration, sleep disturbance, edema, palpitations, cardiac events, kidney issues, elevated liver enzymes. None separated from placebo. A Parkinson's trial at 10 g/day for up to eight years extends that across a multi-year timeline and a broad range of organ-system complaints. No safety differences were found across creatine forms either — the case for monohydrate rests on efficacy and cost, not on safety.
One reading caution travels with this. Creatine Supplementation flags that "no significant difference from placebo" is often framed as causation disproven rather than as insufficient evidence of causation — the distinction developed in Evaluating Evidence & Supplement Quality, and worth holding onto through the rest of this section.
The kidney question has two separable halves. The first is a measurement artifact: creatine converts partly into creatinine, which is the standard biomarker for estimating GFR, so supplementation can make kidney function look worse on routine bloodwork than it actually is. Cystatin C is the more accurate marker in supplement users. The practical protocol is to disclose creatine use to the ordering physician before a metabolic panel, and, if kidney function is already imperfect, to wash out for one to two weeks before the draw. The second half is substantive rather than artifactual — whether excreting the extra creatinine load adds metabolic burden over time — and here the honest statement is about coverage: trials from five days to five years at 5–30 g/day show no measurable effect on GFR or other kidney indices, but they were run in people with healthy kidneys.
That coverage gap is what produces the most interesting dissent in this chapter. A physician source recommends creatine broadly to patients and family and still declines to take it himself, because of a family history of polycystic kidney disease — a slow, usually asymptomatic condition that may not manifest until the 60s or 70s, for which no creatine safety data was identified at all. He treats the absence of data, not evidence of harm, as sufficient grounds for personal caution while judging the population-level risk/benefit favorable, especially for older healthy adults who train, and pursues the same ends through harder training and stricter discipline instead, accepting a slower timeline. It is a clean illustration that population safety data doesn't automatically settle an individual's calculus — the reasoning pattern that Proactive Health & Self-Advocacy takes up more generally.
Two other boundaries: pregnancy safety data is insufficient, so the default stance is to discontinue creatine for the roughly nine-month window as with any non-essential supplement; and women may see more benefit per dose than men, not because the dose differs but because lower baseline stores tied to smaller average muscle mass make the same absolute dose a larger relative change. Creatine is also floated as helpful across periods of rapid hormonal change — menstrual cycle, pregnancy, perimenopause, menopause — with weak evidence for reduced depression in women.
The hair-loss claim is the most instructive single case in this chapter, because the corpus can trace it to its origin. It comes almost entirely from one small randomized trial of 20 college-age male rugby players — 25 g/day for a loading week, then 5 g/day for two more weeks, against placebo. Total testosterone and free testosterone were unchanged. DHT rose 56% after the loading week and stayed about 40% elevated at three weeks.
Three things undercut the inference. The trial never measured hair — no counts, no density, no thickness, only blood hormone levels. The DHT values, despite the percentage jump, stayed inside the normal clinical range throughout: roughly 0.98 → 1.5 → 1.4 against a normal ceiling of 2–3. And it has never been reproduced. A 2021 review of frequently asked creatine questions concluded there is no convincing evidence linking creatine to hair loss and offered regression to the mean as an explanation — the placebo group started with numerically higher baseline DHT (1.26) than the creatine group (0.98), a non-significant gap that could exaggerate the apparent divergence. The mechanism itself is real enough in the abstract: testosterone converts to DHT, DHT binds the androgen receptor in hair follicles and causes miniaturization — but only in people genetically susceptible to male pattern baldness.
As of 2021 the causal question had never been tested directly. It has been since. Creatine Supplementation records a randomized controlled trial published April 2025 in 38 men given 5 g/day creatine or placebo for three months — four times longer than the rugby study — with hair count, density and thickness measured directly alongside hormones. No significant difference between groups in hair parameters, free testosterone, total testosterone or DHT. The corpus describes what appears to be the same trial a second time in more detail: double-blind, 5 g/day creatine against 5 g/day maltodextrin, 12 weeks of resistance training, with DHT flat in both groups at baseline and endpoint. Notably, in both arms alike total testosterone rose and free testosterone fell over the twelve weeks — consistent with the training stimulus itself rather than with anything creatine did.
Layne Norton's summary of the methodological point is the transferable part: "You cannot claim something does X if you don't actually measure X." He states he would revise his position given five more studies showing hair loss, and that the direct evidence currently says the opposite — "There is no evidence to support this and all of the direct evidence we do have says that it does not." One limitation stays on the record: the direct test used a flat 5 g/day and did not replicate the original 25 g/day loading week, so higher-dose loading and longer exposure remain untested. This whole arc — a single small unreplicated study driving a claim that outlives it — is the worked example behind Evaluating Evidence & Supplement Quality.
Creatine's cognitive story is structurally harder than its muscle story. Physical gains are objectively demonstrable inside a twelve-week trial; the cognitive benefit is proposed as conditional — gated by dose on one side and by stress on the other.
The dose problem is anatomical. Muscle acts as a greedy, competitive sink: at around 5 g/day, muscle consumes the supplemented creatine before enough reaches the brain, especially in people who train. On top of that, the blood-brain barrier's astrocyte layer has limited or no expression of the creatine transporter, unlike the endothelial cells on the blood side that do express it — so low blood concentrations from a 5 g dose may not cross effectively, and the mechanism by which higher concentrations eventually do cross is still not well understood. Rhonda Patrick's argument follows from this: a study run at 5 g/day that finds no cognitive effect risks a false negative, because the brain was simply underdosed, and many null brain-effect studies may be exactly that.
The number itself is contested, and part of the confusion is a documented mis-citation. The widely repeated "10 g" brain threshold, attributed to Patrick citing a German study, doesn't match the source: the actual Tübingen study — roughly 30 years old, small n, no control group — used 20 g/day split-dose for four weeks, after which MRI-measured brain creatine rose and then fell back by week 16 post-cessation. A separate placebo-controlled trial in adolescent girls with treatment-resistant depression tested 2 g, 4 g and 10 g over eight weeks; the 10 g arm had the largest raw phosphocreatine increase but did not reach statistical significance against the other doses — consistent with 10 g being an unreliable threshold and 20 g+ being the more robust target. Physionic, reviewing Patrick's Huberman claims, found them essentially correct with the "10 g" figure as the one inaccuracy. Elsewhere in Creatine Supplementation the brain figure is given as ~10 g/day split, so the corpus does not speak with one voice here.
The stress-gating half says dose isn't sufficient either. Per creatine researcher Dr. Darren Candow, via Patrick's podcast, the cognitive benefits concentrate in populations under metabolic or cognitive stress — sleep deprivation, hypoxia, jetlag, aging (generally 60+), neurodegenerative disease — rather than in unstressed people. The rationale is energetic: the brain consumes about 20% of daily energy despite being under 2% of body weight, and creatine taken up by neurons is phosphorylated into phosphocreatine, a reserve buffer that regenerates ATP within seconds of firing. It is the same shape as the muscle case, where mechanical stress is what converts creatine into a strength gain. Physionic flagged the specific meta-analysis Candow cited as methodologically broken — it double-counts the same underlying trials (Alvis, McMorris) by treating multiple measurements from one trial as separate entries, inflating apparent participant counts up to sevenfold and manufacturing a pooled memory effect not replicated in other outcomes like processing speed within the same analysis — while maintaining that the underlying effect is likely real on the strength of corrected analyses and independent RCTs. A 2024 meta-analysis of 16 RCTs likewise found a positive effect on memory.
The sharpest single result is acute rather than chronic: a single 0.35 g/kg dose (roughly 20–30 g depending on bodyweight) given 3–4 hours before cognitive testing in subjects sleep-deprived for 21 hours. The corpus describes its magnitude two ways — in one telling it attenuated but did not fully reverse the declines in brain metabolites, brain pH and test performance; in another it prevented decline on all measured tests, with supplemented sleep-deprived subjects outperforming rested unsupplemented ones on some measures. Either way this is a distinct mechanism from the chronic muscle effect, which needs ~5 g/day over weeks; a single dose does not produce muscle benefits, and a saturation protocol is not what produced this.
Animal work supplies plausible substrate. In brain cells analyzed directly — the invasive measurement humans can't supply — creatine substantially raised coupled respiration, how efficiently Complex V converts ADP into ATP, while leaving maximal respiration unchanged; the effect is specific to ADP-driven conversion efficiency, not a blanket lift in mitochondrial capacity, and ADP is abundant precisely under high demand, including sustained critical thinking. Separately, creatine raised PSD-95 — the postsynaptic protein that anchors AMPA and NMDA receptors to the actin cytoskeleton — by over 50%, while actin itself was unchanged, indicating targeted anchoring machinery rather than nonspecific upregulation. As Evaluating Evidence & Supplement Quality insists, mechanistic animal data supports a conclusion only in combination with human clinical evidence, never on its own. The clinical end of it is thin: a small pilot in Alzheimer's patients at 20 g/day with an apparent placebo control reported improved cognition and, combined with exercise, gains in strength and lean mass, framed inside an "Alzheimer's as an energy crisis" hypothesis. Peter Attia positions creatine-as-ATP-substrate as a candidate for a hypothesized metabolic subtype, started years to decades before onset — a prevention hypothesis, not a proven treatment, and the kind of claim Longevity Interventions is built to classify. How this compares with other cognition-targeted compounds is Brain, Behavior & Recovery's territory.
Patrick's own practice runs ahead of the formal evidence and she says so: she states there is "really no downside" to 10 g/day and predicts it will replace 5 g as the baseline, takes 5 g doses to avoid GI issues, mostly before noon in water or tea, and raises to 15–20 g/day under added stress like post-travel jetlag. She reports avoiding afternoon sleepiness above 5 g and flags that this could be placebo — and that it wouldn't change her behavior.
Creatine is among the few supplements with strong evidence behind it, and — per Jeff Nippard — is routinely overestimated even inside the science-based lifting community. The realistic figure is about 2–3 lbs of lean mass: an 8-week study found +1.1 kg (2.2 lb) against placebo, and a 22-study systematic review in older subjects running from seven weeks to a year found +1.4 kg (~3 lb). A meta-analysis of RCTs puts it at roughly +0.8 kg fat-free mass and −0.28% body fat on average, strengthening when paired with resistance training, with larger effects plausible in vegetarians and vegans given their lower dietary baseline. For scale, a steroid cycle can add 10–20 lbs in a month or two. Real, and modest — the same "temper your expectations, it isn't that it doesn't work" framing Nippard applies to caffeine and to bulking surpluses in Diet Strategy & Body Composition and Resistance Training & Performance Supplements.
Beyond lean mass, the corpus records supporting findings of varying strength: one RCT found creatine lowered HbA1c and post-meal glucose in diabetics; a 2017 review found benefit for injury prevention and recovery; and the Galpin-derived Creatine Monohydrate page adds bone mineral density and cognitive function (memory, executive function) to the list — though, again, that page carries an explicit warning that it was written from model recall and hasn't been through this corpus's normal ingestion, so its specifics aren't anchored. Run through the six-question supplement framework, creatine lands as low-risk, modest-reward: robust safety data, a well-understood mechanism, a real but moderate effect size.
The interesting case is the 2026 12-week resistance-training study in untrained men and women, preceded by a one-week creatine-or-placebo wash-in. Creatine's ~0.5 kg lean-mass advantage appeared entirely during that wash-in week, with no further divergence between groups across the following twelve weeks of training. Both groups gained roughly 2 kg of lean mass overall.
Two readings of that result sit side by side in Creatine Supplementation, and they are not compatible. Norton's is that the benefit is front-loaded and this is what creatine's osmolyte mechanism predicts: water moves into muscle cells, intracellular water reaches a new steady state, and continued training doesn't add further measurable divergence — a baseline-reset pattern he generalizes with a caffeine analogy, and not a debunking, given what he calls mountains of data for lean mass, strength and performance. The Don't Die team's is that the result is a methodology artifact: the seven-day wash-in was meant to control for early water-driven weight gain, but saturation at 5 g/day takes 3–4 weeks, so the study may have taken its post-wash-in baseline before saturation was ever reached — measuring inside the intervention's own onset window rather than after it. Norton independently flags a second limitation pointing the same direction: untrained subjects respond so strongly to any resistance training that the general response may have washed out an ongoing creatine effect, and trained individuals might diverge continuously.
The practical guidance neither reading changes: monohydrate only, 5 g/day, split into a morning and evening half if it upsets your stomach. One disclosure the corpus keeps attached to Norton's commentary — he has a financial conflict of interest, since his supplement line Outwork Nutrition sells a creatine-monohydrate-containing product.