Creatine actually works — which is exactly why you should read the fine print

Most of my job is telling people that the numbers they hear on the news or on podcasts are misleading or not as important as they sound. Supplements are one of the easiest topics for this. Somebody gets excited about a compound, a study gets misquoted, and six months later there’s a $40 tub of it in every kitchen in America.

So it was a bit disorienting to sit down with the literature on creatine and find that the hype is actually attached to something real. Well, mostly real. Let me show you where the seams are.

Start with what creatine was actually invented to do. A meta-analysis out this year in Frontiers in Nutrition pooled eleven randomized trials of over-60s who lifted weights, half of them also taking creatine. The creatine group got stronger — significantly stronger, on average, than the people doing identical training without it.

Granted, the effect size of 0.31 sounds more impressive than it is until you translate it. An effect size of 0.31 means your average creatine-taker finished stronger than roughly 62 percent of the placebo group. Turn that around, and 38 out of every 100 people who took nothing still beat the average person in the study taking creatine. So that is a real edge, and I’d take it. It is also not what is being sold to you on Instagram today.

Now here’s the study that should make you distrust your own bathroom scale: A team at the University of New South Wales published a trial last year in Nutrients . They supervised 63 adults — half on five grams of creatine a day and half on placebo — through 12 weeks of supervised lifting. However, they also built in a seven-day “wash-in” before anybody touched a weight, and they scanned everyone with DEXA at every stage.

After that first week of creatine and precisely zero exercise, the supplement group had put on half a kilo of what the scanner read as “lean body mass.” But muscle doesn’t materialize in seven days. Rather, that was just water being pulled into muscle cells, which the DEXA scanner obligingly counts as lean tissue. Then, over the 12 weeks of actual training that followed, both groups put on about two kilos, and the gap between them went away — same training, same result whether or not you were paying for the powder.

I keep coming back to that study because of what it explains. Everyone who has ever started creatine, weighed themselves five days later and felt enormous was not making it up. They measured a real thing at the wrong moment and drew the wrong conclusion from it. What I would like you to consider is that the industry has been quietly selling that first week for 30 years. The instrument and the customer are both fooled, even though the whole thing is sincere enough that nobody involved is lying about anything.

Then there’s the brain business. Sometime around 2024, creatine started being marketed as a cognition- and memory-enhancing substance rather than a gym aid. One industry analysis counted 469 podcast episodes on creatine in 2025 alone — more than the entire preceding decade put together — with brain-focused framing in the titles nearly doubling in two years and one title in six telling the listener she needs to be taking this.

The evidence underneath all that noise is thin. A 2024 review pooled sixteen randomized trials on creatine and cognition. The total number of participants in all of these was only 492 people, spread across ages 20 to 76, on different doses, for different durations, tested on different instruments. The review found small improvements in memory and attention and no statistically significant effect on overall cognitive function or executive function.

And there is a recurring statistical problem in this corner of the literature that I think deserves naming. When an earlier meta-analysis in Nutrition Reviews reported a memory benefit, a letter to the same journal pointed out that several outcomes measured on the same participants had been treated as if they were independent observations — in other words, they were double-counting the benefits. This narrows the confidence interval artificially and makes a wobbly result look settled. It is a boring, technical error, but it makes a difference and it is everywhere.

Break those same memory results out by age, and the story falls apart entirely. For adults 66 to 76, the effect size is genuinely large, but the effect is negligible for ages 11 to 31. So a fair reading would be that creatine might do something for an older brain running short on metabolic headroom, but it does approximately nothing measurable for a well-rested 27 year-old — you know, the guy who is actually buying all that creatine.

In fairness, I should also point out that the panic over safety of creatine has badly outrun the data. Creatine does not hurt healthy kidneys. The International Society of Sports Nutrition says so. A meta-analysis of renal function in athletes says so. And a full season of monitoring in 71 female soccer players turned up nothing beyond clinically trivial wobble in the bloodwork.

Creatine does raise serum creatinine (note the extra two letters in the spelling), which happens to be the metabolite your doctor uses as a proxy for kidney function. So your lab report might look frightening after you take creatine, even if nothing is actually wrong with your kidneys. If you have kidney disease already, ask your doctor before taking creatine. If you haven’t, then tell him you are taking it before he tests you and panics over your labs.

I want to finish with the study that keeps me cautious, even as I’m being persuaded. The biggest and longest trial in the field put 237 postmenopausal women on about 10 grams a day for two years, alongside supervised resistance training and walking. The primary endpoint was femoral neck bone density. The result was null — no effect on bone, bench press or hack squat. Creatine did produce a faster 80-metre walk, about a second and a half, and that finding is solid.

The lean tissue gain you will see quoted everywhere came out of a sub-analysis of valid completers rather than the randomized sample, and only 174 of the 237 women finished. So that is, at best, encouraging and inconclusive. But that is not how you will see it described on a label.

Five grams a day of creatine monohydrate is cheap, better studied than nearly anything else on the supplement shelf, and possibly useful for some people. If you are over 60 and lifting weights, then I think it is not a terrible idea, and I’d say the same to my own mother.

But look at what every single one of those findings has in common: The creatine did something for people who were already training. The barbell is the intervention — the powder is just a rounding adjustment.

So buy it if you want, but remember that it’s the weight-lifting that’s important, not the supplement.

Dr. Liberty Vittert Capito is a professor of the practice of data science at Washington University in St. Louis.