The Creatine Loading Phase: What the Founding Trials Actually Measured

Almost every creatine label carries the same two-stage instruction (one exception, a 2025 label that lists the 20 g a day with no end date, is set out on the Orgain creatine page), and almost every explanation of it is written in the language of performance: load to get the benefit sooner. The papers underneath it are about something narrower and more interesting. They measured a metabolite in a muscle, in a small number of men, and one of them contains a result that the marketing does not mention. This is research journalism and contains no usage guidance, per our editorial standards.

Where the numbers came from

Two papers, 48 male subjects between them, both using needle biopsy of the quadriceps femoris.

That last row is the one worth sitting with. The loading phase is described everywhere in terms of strength, power and how quickly a training block starts working. The two studies that produced its arithmetic measured neither strength nor power nor muscle size. They measured how much creatine was in the muscle (Harris et al. 1992, PMID 1327657; Hultman et al. 1996, PMID 8828669).

The result the label leaves out

Hultman and colleagues did not only test the fast route. In the same 31 subjects they ran the slow one, and reported it in the abstract's own words: 20 g a day for six days raised muscle total creatine by approximately 20%, and "a similar, but more gradual, 20% increase in muscle total creatine concentration was observed over a period of 28 days when supplementation was undertaken at a rate of 3 g/day."

Their conclusion states the consequence directly — that 3 g of creatine a day "is in the long term likely to be as effective at raising tissue levels as this higher dose."

So the study that established loading also established that loading is a shortcut rather than a requirement. Both routes arrive at roughly the same 20% rise in muscle total creatine; one takes six days and one takes four weeks. Every subsequent claim that loading produces a larger effect, rather than an earlier one, is unsupported by the paper it is usually credited to. Some brand directions now go past it: one store's FAQ describes six 5 g servings a day for six days, half as much again as the 1996 protocol (Naked creatine).

What a loading dose actually does once swallowed

Harris and colleagues tracked the fate of the dose, and the numbers are less flattering than the protocol suggests.

• Uptake into muscle was greatest during the first two days, accounting for 32% of the dose administered in the three subjects taking 6 × 5 g a day.
• Renal excretion over the first three days was 40%, 61% and 68% of the dose.
• Roughly 20% or more of the creatine taken up was measured as phosphocreatine.
• No change was apparent in muscle ATP content.

The declining retention across those three days is the entire mechanical case for a short window: the muscle takes up most of what it is going to take up early, and the rest increasingly goes to the kidneys. It is also why the maintenance figure is so much smaller than the loading figure — 2 g a day was enough to hold the elevated concentration for a further 30 days in the 1996 study. How the loading amount is split across the day has its own gut-tolerance trial, set out on why a large single creatine serving upsets some stomachs.

Harris 1992 also supplies the most quotable line in the creatine literature, and it is a unit conversion rather than a claim: a single 5 g dose "corresponds to the creatine content of 1.1 kg of fresh, uncooked steak." The same sequence — muscle first, then the scale, then performance — is laid out day by day on how long creatine takes to work.

Who responds most

Not everyone moved the same distance. Harris and colleagues reported that the increase "was greatest in subjects with a low initial total creatine content and the effect was to raise the content in these subjects closer to the upper limit of the normal range. In some the increase was as much as 50%."

That is a ceiling effect, and it has a practical consequence the trials never tested directly: a person whose muscle creatine already sits near the top of the normal range has less room to gain, however much is administered. Diet is the obvious reason for a low starting point, which is why the steak conversion is more than a curiosity.

What happens when it stops

Thirty days, roughly. In Hultman 1996, once the 2 g/day maintenance intake was withdrawn, muscle total creatine "gradually declined, such that 30 days after the cessation of supplementation the concentration was no different from the presupplementation value," with urinary creatinine excretion correspondingly increased across that window.

There is no rebound below baseline in the data, and no withdrawal effect reported. The muscle returns to where it started. Two later biopsy studies timed the same washout at under four weeks; they, and what the drain means for cycling, are on what happens when you stop creatine.

Has anything since compared loading against no loading?

Very little, and the one recent trial that ran the comparison directly found no separation between the two monohydrate arms. Eghbali, Arazi and Suzuki randomised 40 participants aged 18–25 into four groups of ten for eight weeks alongside resistance training: creatine hydrochloride, creatine monohydrate with a five-day loading phase at 0.3 g/kg followed by 0.03 g/kg, creatine monohydrate without loading at 0.03 g/kg, and placebo.

All three supplement groups improved strength, arm and thigh cross-sectional area and skeletal muscle mass significantly against placebo. The paper reports the supplement groups together and does not report a loading-specific advantage over the non-loading monohydrate arm (Eghbali et al., Physiological Research 2024, PMID 39545789).

Ten people per arm over eight weeks is a small test, and it should be read as one, not as a settlement. The honest summary of the comparative evidence is that it is thin: the question of whether loading changes an eight-week training outcome has essentially not been asked at a scale that could answer it.

That trial is also the one place in the literature where two different salts were given at the same amount per kilogram, which makes it the best available test of a separate marketing claim — whether the hydrochloride form does anything monohydrate does not.

Two of the trials that loaded at these amounts also gave caffeine, and what happened then is set out in the interference literature.

The limits of this page

Two studies from 1992 and 1996, 48 male subjects, one biopsied muscle, one metabolite. That is the foundation of a protocol printed on tens of millions of containers. It is also worth noting that the one biopsy-confirmed case of kidney injury in this literature followed a six-day high-dose loading regimen rather than ordinary maintenance intake — the case, and the trials that measured filtration properly, are on our page covering creatine and the kidneys. The findings have held up — they have been replicated in the sense that nobody has overturned them — but they were never large, never mixed-sex, and never about performance.

What the evidence supports is narrow and worth stating plainly: creatine intake raises muscle total creatine; a higher intake raises it faster; a smaller intake raises it eventually; a small maintenance intake holds it; stopping returns it to baseline in about a month. The reason a faster route cannot produce a fuller store is a hard storage ceiling of about 160 mmol per kilogram of dry muscle, which we set out in the arithmetic behind the muscle creatine pool — and the surplus that ceiling turns away is measurable in urinary creatinine, the marker separated from its two similarly named neighbours on creatine in bloodwork. The 20 g and 2 g figures here are only two of at least three amounts in general circulation, and where each of those numbers came from is a separate question from which is correct. That slow turnover is also why the hour at which it is swallowed has never produced a measurable difference, and why the water it draws in goes where the compartment measurements say it goes. Everything beyond that, including the effect on training outcomes, belongs to the much larger body of resistance-training trials and carries a small effect size when it is measured directly.

Sources and dates

All searches run 2026-09-11 on PubMed E-utilities. A control query (minoxidil AND alopecia AND "randomized controlled trial"[pt]) returned 149 records, reproducing this site's previous count.

Figures are reported as the sources state them. Nothing here is a recommendation, a schedule or an instruction; the compounds and amounts described are reported, not advised.