Creatine Benefits: They Screened 250 Supplements. Two Survived.

In 2003 two researchers did something the supplement industry has never quite recovered from. They took every dietary supplement claimed to build muscle, set a minimum standard of evidence, and counted how many cleared it.

They examined 250 supplements. Only six had more than two studies that met the entry criteria — three weeks minimum, resistance training at least twice a week. [1]

Of those six, two produced a significant effect.

Creatine was one of them. Twenty-odd years and several meta-analyses later, it still is.

How much lean mass, actually

The most carefully run analysis we found, preregistered and published in 2024, pooled twelve trials in adults under fifty, with the protocol registered in advance so the authors could not go fishing after the fact. [2]

Against resistance training alone, adding creatine produced 1.14 kg — about 2.5 lb — of extra lean body mass (95% confidence interval 0.69 to 1.59 kg). Body fat mass fell by 0.73 kg, roughly 1.6 lb, and body fat percentage by 0.88 points. [2] Those twelve trials were almost entirely men (13% women) and mostly experienced lifters, and the authors say plainly that the number cannot be generalised to women; the older-adult trials below included both sexes.

The authors put it in practical terms: about 7 grams a day, or 0.3 grams per kilogram of bodyweight, “is likely to increase LBM by 1 kg and reduce fat mass by 0.7 kg more than RT alone.” [2]

Two and a half pounds is not a transformation. It is also two and a half pounds you would not otherwise have, from a compound that costs pennies a day and has been studied for thirty years. Almost nothing else in the supplement aisle can make that claim — which is rather the point of the table above. One caution before you take it to the mirror: that is lean mass as a body scan measures it, and a scan counts water as lean. How much of it is muscle is a separate question, and it is taken up under “before and after” below.

The finding that runs against the marketing

Creatine is sold to young men. It appears to work at least as well in people considerably older.

A meta-analysis of 22 trials covering 721 participants with a mean age between 57 and 70, training two to three days a week for anything from seven weeks to a year, found creatine added 1.37 kg of lean tissue — about 3 lb (95% CI 0.97 to 1.76, p < 0.00001). [3]

Strength improved too: standardised effect sizes of 0.35 for chest press and 0.24 for leg press.

Compare that with the 1.14 kg in the under-fifties. The older group’s number is a little higher, but the two come from different meta-analyses, their ranges overlap, and the older-adult trials ran far longer (seven weeks to a year, against an average of about eight weeks), so more is not something the evidence establishes. What it does establish is that creatine works at all in people the marketing never addresses. The idea that people who have lost muscle with age have more room to top up is the desk’s surmise, not a finding.

What it actually does, in one paragraph

Your muscles hold a small store of phosphocreatine, which regenerates ATP — the molecule that powers a contraction — during very short, very hard efforts. That store runs out in seconds, which is why the eighth rep is harder than the second.

Supplementing raises the size of that store. The International Society of Sports Nutrition’s position stand puts it plainly: creatine “increases intramuscular creatine concentrations which may help explain the observed improvements in high intensity exercise performance leading to greater training adaptations.” [5]

Note the causal chain, because it matters. The route the position stand describes is indirect: creatine lets you do slightly more work, and the extra work builds the muscle. That is the usual explanation, and it is why four of the six pooled analyses on this page counted only trials in which people lifted [1] [2] [3] [9]. But it is an explanation, not a finding. The twelve-trial analysis tested it and found that the amount of training people did made no difference to the size of creatine’s edge; its authors concluded the extra lean mass was “likely because of other mechanisms” and named body water as a candidate [2]. And the 100-study analysis in the next section was never limited to training studies. It searched for creatine supplementation of any kind, and its abstract, the published summary, reports a larger effect on body composition after loading-only courses, the short high-dose kind, than after longer maintenance dosing [4]. We could read only the abstract, and it does not say how many of the 100 studies included a training programme. What creatine does with no training at all is taken up under “before and after” below.

Where the effect is biggest, and where it is smallest

The largest early meta-analysis pooled 100 randomised, placebo-controlled studies and broke the results down by the kind of exercise. [4]

Effect sizes — a standard scale where 0.2 counts as small, 0.5 moderate and 0.8 large — came out at 0.42 for upper-body exercise, 0.21 for lower body, and 0.13 for whole-body work. Laboratory tasks like isometric or isokinetic testing showed larger effects (0.25) than field tasks like running or swimming (0.14). [4]

Two honest observations follow. Creatine helps most with short, hard, repeated efforts — which is what a set of bench press is, and is not what a five-kilometre run is. And every one of those numbers is small on the conventional scale.

That is worth sitting with, because it is why we rate creatine Established rather than remarkable. Established means the evidence is consistent and the direction is not in doubt. It does not mean the effect is large.

A caveat on that table, because a date matters

Look back at the 2003 comparison and you will see protein listed as having no significant effect on lean mass or strength. [1]

That was 2003. Protein’s evidence base has moved a very long way since, and it would be dishonest to leave a twenty-three-year-old null standing as a current verdict. We flag it here rather than quietly cropping the row, because cropping inconvenient rows is what this site exists to catch.

What the 2003 result does tell you is how thin the evidence for everything else was, even then. Chromium, DHEA and androstenedione were all being sold hard at the time. None of them cleared the bar.

About “before and after”

This is one of the most searched questions about creatine, so here is the honest version.

Some of what you see in the first fortnight is water. The position stand quoted above describes creatine loading bringing on a short-term gain of about two to four cups (0.5–1.0 L) of body water, roughly in step with the early rise on the scale [5]. Where that water ends up is less settled than most articles say, and than this page used to say: we told you it is drawn into the muscle cell, and the studies that have measured it do not establish that. Either way, water is not muscle tissue. And the lean mass numbers above do not sit on top of that water, which is what this page said until now. They include it. We covered the weight question, and what the water studies found, in its own verdict.

Here is why. A DXA scan, the low-dose X-ray body scan these trials used, cannot tell water from muscle: lean mass, on the scanner, is everything that is not fat or bone, body water included [9]. And in the twelve trials behind the 2.5 lb figure, the first scan came before the first dose. The authors’ own data sheet records it that way for eleven of the twelve and as not reported for the other, and ten of the twelve then took a loading dose of about 20 grams a day for five to seven days [2]. So whatever creatine does to body water in that first week is inside the 2.5 lb, not underneath it.

So how much of it is water? In 2025 the group behind that twelve-trial analysis published a trial set up to separate the two [6]. They scanned 63 untrained adults, a little over half of them women. One group then took 5 grams of creatine a day for a week and did no training; the other took nothing. Everyone was scanned again. The creatine group’s lean mass was up 1.1 lb (0.51 kg). The other group’s had not moved. Then both groups lifted three times a week for twelve weeks, the creatine group still taking it, and the 54 who finished gained about the same: 4.9 lb (2.24 kg) with creatine, 4.7 lb (2.11 kg) without. A gap that small turns up by chance about seven times in ten (p = 0.71). Measured the way the older trials measured it, from the very first scan to the last, the creatine group did finish ahead, 6.1 lb (2.78 kg) against 4.5 lb (2.04 kg), right on the line researchers use for “probably not chance” (p = 0.05). So this trial agrees with the older ones about the total. What it adds is when the lead appeared: most of it in the first week, before anyone had touched a weight.

One trial does not overturn twelve, and this one has limits its authors set out. The comparison group took nothing, not a dummy powder, so the participants knew which group they were in; the people reading the scans did not. The first-week gain was smaller than the scanner’s own error from one scan to the next, though, as the authors point out, that error falls on both groups alike. It showed up in the women and not the men when the two were separated afterwards, and in the trunk, not the arms or legs. The trial did not measure body water, so “it was water” is the authors’ reading of the result, not a measurement. The dose was 5 grams with no loading week, and the authors’ own conclusion is that a larger one might do more. The creatine was supplied by a supplement company.

There is a more direct way to ask: measure the muscle, not the whole body. A 2023 analysis pooled ten trials that did, nearly all by ultrasound, in people who trained for anything from six weeks to a year [9]. Creatine’s edge over training alone came to about a twentieth of an inch of muscle thickness (0.10 to 0.16 cm, depending on the muscle). The authors call that trivial to small. They put the chance that there is a real edge at all at about 96 in 100, and they offer body water as one reason the scanner figures run so much larger.

And with no training at all? Less is known than the usual story suggests. Pooled, the trials in which people took creatine and did not train show no clear gain in lean mass (the analysis calls it fat-free mass, everything that is not fat): 0.5 lb (0.24 kg) across seven comparisons, with a range of plausible answers running from a loss of 0.6 lb to a gain of 1.7 lb (−0.26 to 0.75 kg), against 2.2 lb (0.99 kg) where people lifted [7]. That analysis’s own test of whether its exercise groups differed did not reach the usual bar (p = 0.14). Individual trials have found a gain without training. One is the first week of the 2025 trial above. Another, from 2026, gave 10 grams a day or a dummy powder for twelve weeks to 26 adults of 45 to 65 who had chosen not to join an exercise programme: lean mass rose 2.4 lb (1.1 kg; the range of plausible answers runs from 0.4 to 4.4 lb, or 0.2 to 2.0 kg) on creatine and did not move on the dummy [8]. In that trial the creatine came from its manufacturer, the corresponding author chairs that manufacturer’s scientific advisory board (both are declared in the paper), and the trial was registered after it had begun. Neither trial measured body water. So what they show is more lean mass on a scanner without training. Whether any of it is muscle, they cannot say.

Two and a half pounds of lean mass over a training block does not photograph dramatically. Anyone showing you a picture that does is showing you training, lighting, and time — with creatine somewhere in the mix.

The verdict

Creatine, taken while you train, adds lean mass and strength. That is Established — replicated across independent meta-analyses in young adults and older adults, in a hundred randomised placebo-controlled studies, with a known dose and the same direction in every pooled analysis on this page. What that rating covers is more lean mass on a scanner and more strength. How much of the scanner figure is muscle is not settled.

Expect roughly 2.5 lb of extra lean mass in adults under fifty — a figure drawn mostly from trained men — and a similar 3 lb in older adults of both sexes, on top of what the training itself delivers. It works best for short, hard, repeated efforts. Every one of those figures comes from people who trained. Without training, pooled trials show no clear gain in lean mass, with a range too wide to rule a small one out [7], and the two recent trials that found one did not measure body water, so they cannot say it was muscle [6] [8].

The reason it stands out is not that the effect is big. It is that in a field where 250 candidate supplements were screened and only two cleared a modest evidence bar, this one keeps clearing it — and it is one of very few things in this whole corpus that the supplement aisle sells and the evidence actually supports.

The rest of what you probably want to know is already covered: how much to take, whether it is safe, the weight gain, and the hair loss question.

Related: is creatine safe · how much creatine should you take · do testosterone boosters work · fish oil benefits