Creatine: The Kick, Not the Engine

Supplements
Written By Cat Hoh

Understanding how creatine supports performance, who it benefits, and where it fits in an athlete's program.

Creatine is the most studied supplement in sport and one of the few that reliably does what the tub says it does. Distance runners have largely ignored it, on the sensible grounds that it is a strength supplement and they are not strength athletes. That instinct is half right, and the half that is wrong is the half that decides races. Creatine will not raise your aerobic ceiling. The best evidence on that point is unambiguous, and it points the wrong way. What creatine may do is sharpen the ninety seconds that separate fourth from first.

The other engine, the one you are not training

Almost everything a distance runner does in training is aimed at the aerobic system, and rightly so. But races are not run at the pace the physiology textbook assumes. They contain a move at 3km, a hill someone attacks, a pack that surges when the leader stops covering, and a final lap where whoever has anything left uses all of it. Those moments do not run on oxygen kinetics. They run on the phosphagen system, and that system is the one creatine actually touches.

The mechanism is not controversial. Supplementation raises intramuscular phosphocreatine, the substrate for rapid ATP resynthesis when demand spikes faster than aerobic metabolism can answer. Forbes and colleagues set out the plausible pathways: faster ATP turnover, some buffering of hydrogen ion accumulation, better glycogen resynthesis when creatine is taken with carbohydrate, lower inflammation and oxidative stress, and possibly a nudge to mitochondrial biogenesis. Muscle biopsy work backs the first link in that chain. Roberts and colleagues measured phosphocreatine up around 16% against placebo after six days of loading.

Raising the substrate is not the same as improving the performance, and this is where the enthusiasm gets ahead of the data. Finn and colleagues loaded triathletes, confirmed their intramuscular creatine had gone up, and then found no ergogenic effect at all across four repeated twenty second cycle sprints. The tank filled. Nothing happened.

The benefit has a clock running on it

What separates the trials where creatine works from the trials where it does nothing turns out to be duration, and one dataset shows it more cleanly than any argument. McNaughton tested kayak paddlers over three efforts of different lengths and measured total work done against placebo.

Figure 01 / Interactive

The longer the effort, the less creatine has to offer

Effort duration / tap a stop to change

90 s 150 s 300 s 90 second test 150 second test 300 second test
+16.2% +13.6% +6.63% 90 s 150 s 300 s

+16.2% more work than placebo. Short enough that phosphocreatine carries a large share of the load. Reported as statistically significant.

+13.6%, and no longer significant. The advantage is shrinking and the certainty is going with it.

+6.63%, under half the 90 second effect. Five minutes in, the aerobic system is running the race and creatine is a bystander.

McNaughton et al. (1998) in kayak paddlers, as reported in Wesółowski et al. (2026). Reported values, drawn to scale. Read the slope, not the individual bars: the 90 and 300 second comparisons reached significance while the 150 second one sitting between them did not, which is what small samples do to an otherwise orderly trend.

As phosphagen-dependent energy turnover gives way to oxidative metabolism, the effect decays. That is the whole field compressed into one figure, and it explains why creatine looks like a miracle in a repeated sprint protocol and like an inert powder in a time trial.

What it will not do for you

Here is the part the marketing leaves out. Gras and colleagues pooled nineteen randomised controlled trials to ask whether creatine improves VO2max. It does not. VO2max rose in both arms, because most participants were training, but it rose less in the creatine groups than in the placebo groups.

Before concluding that creatine is bad for your aerobic system, look at what VO2max actually is. It is oxygen uptake divided by body mass. Creatine draws water into muscle and body mass goes up, typically between 0.6 and 2.5% on a standard loading protocol. Leave the numerator untouched, inflate the denominator, and the number on the lab report falls.

Figure 02 / Interactive

Nothing changes in the engine. The number still drops.

Body mass gained on loading / tap a stop to change

None +0.6% +1.5% +2.5% No mass gain Mass gain 0.6 per cent Mass gain 1.5 per cent Mass gain 2.5 per cent
70.0ml/kg/min

Our 70 kg runner at baseline, carrying nothing extra. Absolute oxygen uptake is 4.90 L/min and stays there at every stop.

69.6ml/kg/min

Down 0.4 at the low end of the reported range. Absolute uptake unchanged at 4.90 L/min. Now carrying 420 g, roughly a full drink bottle, on every stride.

69.0ml/kg/min

Down 1.0, which most coaches would read as a bad block of training. Absolute uptake unchanged at 4.90 L/min. Now carrying 1,050 g for 25 laps.

68.3ml/kg/min

Down 1.7, or 2.4%, with the engine exactly as it was. Now carrying 1,750 g, and it is not going anywhere.

Arithmetic, not a study finding. The mass range is the one reported across trials in Wesółowski et al. (2026); the 1.5% stop is an illustrative midpoint rather than a measured condition. This does not prove mass caused the result in Gras et al. (2023), which reported a standardised effect size and not a percentage. It shows why a null on this particular metric was close to inevitable.

For a runner the denominator is not only on paper. You carry it. A rower or a cyclist can absorb two kilograms of intracellular water and barely notice. Someone racing 10,000m is paying for every gram of it, on every stride, for twenty-five laps. This is the trade the whole decision turns on, and no study has properly priced it.

Creatine supplementation increases body mass, which may offset the potential positive effects, particularly in weight-bearing activities. Forbes et al., Journal of the International Society of Sports Nutrition, 2023

Why everyone who takes it swears it works

There is a second case for creatine in endurance athletes, that it blunts the inflammation of hard training. Three trials found lower markers after severe efforts, including runners after a 30km race, and the effects were not small. But the muscle damage markers go in every direction at once, and two studies in footballers three years apart disagree about whether creatine raises or lowers creatine kinase. A separate meta-analysis found no consistent reduction in CRP or IL-6 at all. Three studies pointing one way is a hypothesis, not a finding.

The more useful thing to understand is why creatine feels so convincing to the people taking it. Here one study does the job better than any argument. Wang and colleagues reported both numbers for the same athletes on the same tests: what improved from baseline, and what actually beat placebo.

Figure 03 / Interactive

The number athletes quote, and the number that counts

Outcome measured / tap to change

Half-squat 1RM Jump height Peak power output
+33.4%+7.65% +19.7%+1.24% +11.7%−2.2% FROM BASELINE VS PLACEBO

A third stronger. The rest was training. Against placebo the gain shrinks to under a quarter, and even that was descriptive only.

A fifth more jump height. It would feel like the supplement was working. Against placebo the effect all but vanishes, and is not significant.

Peak power up, every athlete better. Against placebo it points backwards. The placebo group improved more.

Wang et al. (2018), in baseball, basketball and tchoukball players, as reported in Wesółowski et al. (2026). Reported values, drawn to scale. These are mixed-sport athletes and strength outcomes, not distance runners. The lesson is about how to read a supplement study, not about transfer to endurance. The only trial in this review conducted exclusively in women shows the same trap: swimmers' drag improved 16.2% from baseline and came out significantly worse than placebo.

Everyone in that study got stronger, and the supplement deserved almost none of the credit. This is the shape of nearly every creatine anecdote you will hear. An athlete starts creatine alongside a training block they were doing anyway, gets better, and attributes it. The placebo group also got better, which nobody in the real world ever gets to see. When an athlete or a rep or a podcast hands you a percentage, the only question worth asking is which of those two columns it came from.

How much, and what that looks like on a plate

If you decide it is worth trying, the protocols are settled enough to be boring. Kreider and Stout put the fastest method at 5g of creatine monohydrate four times a day for five to seven days, about 0.3g per kg, which is the 20g daily loading dose used in most of the endurance trials. After that, 3 to 5g a day holds the stores. Skip loading and 2 to 3g a day gets you there in about a month, more slowly and with less fanfare. Carbohydrate matters more than most people realise: the ISSN position stand notes that adding 93g of carbohydrate to 5g of creatine raised total muscle creatine by 60%, and that stores drift back to baseline four to six weeks after you stop.

How much you gain depends on where you start. Someone with low stores, which usually means someone who eats little meat or fish, can add 20 to 40%. Someone already close to saturation gets 10 to 20%. Vegetarians run 20 to 30% below omnivores, so the athlete most likely to respond is the one whose diet has already been doing the opposite of loading for years.

Which raises the obvious question. Creatine is a normal nutrient, present in ordinary food. Why not eat it? Your body makes roughly half of what it needs and the rest comes from red meat and fish, but across 89,161 people in the NHANES database the average American actually eats 0.70g of creatine a day. Meat and fish average 3.88g per raw kilogram. Do the division.

Figure 04 / Interactive

What the dose looks like as dinner

Daily creatine target / tap a stop to change

0.7 g 3 g 5 g 20 g 0.7 grams, average diet 3 grams, maintenance 5 grams, study dose 20 grams, loading dose
180 g 770 g 1.3 kg 5.2 kg AVERAGE DIET MAINTENANCE STUDY DOSE LOADING RAW MEAT OR FISH

180 g of meat or fish. One modest chicken breast. This is what the average person already eats, and it is a fifth of the dose every trial in this article used.

770 g a day, every day. The low maintenance target. Achievable, at roughly a kilo of steak every two days, forever.

1.3 kg a day. The dose in the studies. Three large steaks, daily, with the calories and saturated fat that ride along.

5.2 kg a day, for a week. The loading dose. This is not a diet. It is why the supplement exists.

Calculated from the 3.88g per kilogram average across creatine-containing foods used by Korovljev et al. (2021), whose 89,161-person NHANES analysis also gives the 0.70g average intake. Doses from Kreider & Stout (2021). These are raw weights and a cross-food average. Real content varies by species, with oily fish such as herring at the top, poultry lower, dairy negligible and plants essentially zero. Treat the bars as orders of magnitude, not a shopping list.

That is the honest answer to the food question, and it is worth giving to any athlete who would rather eat than supplement. You cannot eat your way to a loading dose. You can eat your way to a maintenance dose if you are willing to build your diet around it. And the athletes with the most to gain from supplementing are precisely the ones least likely to want to.

Who has actually been tested, and who is doing the telling

The 2026 scoping review pulled together thirty-eight randomised trials spanning three decades, which sounds substantial until you look at who was in them. Twenty-five enrolled men only. Exactly one studied women exclusively. Samples ran from eight participants to forty. Swimming was the most studied discipline, with cycling and rowing next, and distance running barely features. A meaningful share of participants were recreationally active or untrained rather than competitive athletes, and almost nothing was done in a real race.

Then there is the question of who is making the case. The Forbes review is a fine piece of work and its mechanistic account holds up, but it appears in the journal of the International Society of Sports Nutrition, two of its authors are that society's chief executive and vice-president, the society takes sponsorship from companies that sell creatine, and one author sits on the scientific advisory board of a creatine manufacturer and previously served as chief scientific officer of a company selling creatine products. The dosing review cited above is in the same position: its publication costs were paid by a creatine manufacturer whose scientific advisory board its lead author chairs. All of it is disclosed, properly and in full, which is the system working. It still means the most enthusiastic reading of this literature and the commercial interest in creatine are not at arm's length.

Contrast that with the independent syntheses. Gras found no VO2max benefit. Fernández-Landa and colleagues, pooling trained populations, concluded creatine does not consistently improve endurance performance broadly. And a 2023 meta-analysis in Cureus repays reading for a different reason: its abstract reports that creatine beat placebo, while its own pooled forest plot favours placebo with an odds ratio of 2.51 and its funnel plot shows publication bias across the three studies analysed. A paper that contradicts itself between its results and its summary is not evidence for either position. It is a warning about how thin the ground is.

So what do you tell the athlete in front of you

Match the supplement to the demand, not to the sport's label. If the event is decided by surges, a finishing kick, repeated accelerations off a pack, or a track cyclist's last two hundred metres, creatine is defensible and the mechanism supports it. If the event is a steady-state effort where the winner is whoever holds a pace nobody else can, the case thins to almost nothing and the added mass starts working against you before the start line.

Practically: weigh the athlete before and after loading, because a two per cent gain in a weight-sensitive runner is a decision point and not a footnote. Take it with carbohydrate. Give any trial a full training block before judging it, since five days of data will tell you nothing that is not noise. And be honest about what you are buying: not a bigger engine, just a slightly sharper one at the moments the engine is not what matters.

One thing sits unresolved, and it should. The entire case rests on studies where creatine's benefit was measured in kayak ergometers, cycling sprints and swim sets, and its cost, the mass, falls hardest on runners, who have barely been studied at all. The one study here that came closest to a real distance running event measured inflammatory markers, not finishing times. Nobody has yet run the trial that would answer the question, and until somebody does, everyone recommending creatine to distance runners, this article included, is reasoning by analogy from a boat.

References

Cat Hoh
Next

From Carbon to Track