Carbohydrate mouth rinsing: the effect is real, but smaller than the hype suggests
A meta-analysis of 35 studies confirms that rinsing your mouth with a maltodextrin solution does help — but the effect is small (g = 0.18), the certainty of evidence is low, and the classic “only works when fasted” rule has barely been tested at all.
There is something instantly appealing about this trick: you take a mouthful of sports drink, swill it around for a few seconds, spit it out — and ride faster. No calories, no load on the stomach. It sounds like a physiological magic trick, and on paper it really does work differently from ordinary fuelling. The only question is the size of the effect.
How can this work at all
An ordinary gel or drink has to be absorbed, reach the bloodstream and get to the muscles — that takes time and a functioning gastrointestinal tract. With a rinse none of that happens: the solution never enters the body.
The mechanism sits higher up. The oral cavity contains receptors that respond to carbohydrate, and they are linked to brain regions responsible for motor control and motivation. The brain gets a “fuel is on the way” signal and, it seems, slightly relaxes its own limits on power output. The effect is central rather than metabolic — which is why it appears almost instantly.
Hence the elegant hypothesis that has circulated through coaching articles for years: if the signal is about fuel, then it should work best when fasted, when the body genuinely has something to worry about. The recent meta-analysis is titled, pointedly, “Fed, not fasted”. And this is where it gets interesting.
What the data showed
A three-level meta-analysis (2025) pooled 35 studies, 76 effect size estimates and 444 participants (380 men and 64 women).
The overall result: the effect exists, but is small — Hedges' g = 0.18 (95% CI 0.09–0.28, p < 0.01). For reference, 0.2 is conventionally the lower bound of a “small” effect.
And what matters more than the estimate itself are the two numbers next to it:
- Heterogeneity I² = 65% — the studies diverge substantially from one another.
- Prediction interval: −0.30 to 0.66. This is the crucial one. It means that in a new, not-yet-conducted study the result could well be negative. The mean effect is positive, but reproducibility is weak.
By protocol the picture is clearer:
- A rinse duration of ≤10 seconds gave a consistent benefit; the 5-second version — g = 0.25 (p < 0.01).
- A concentration of around 6.4% proved sufficient, and no “more sugar, more effect” relationship was found.
- Maltodextrin worked better (g = 0.21, p < 0.05), sucrose — barely at all. That makes sense: sweetness and activation of carbohydrate receptors are not the same thing, and maltodextrin is essentially not sweet.
And what about the central “fasted” hypothesis? It turns out there are only four direct comparisons of fed versus fasted states, and they contradict one another: in one study the best result came from combining the rinse with a high-carbohydrate pre-exercise meal, in another the benefit appeared only after glycogen depletion, in a third the effect was independent of feeding, and in a fourth there was no effect at all. In other words, the famous rule that it “only works on an empty stomach” has essentially not been tested.
How to apply it
Mouth rinsing is a narrow tool, not a replacement for fuelling.
- Where it fits: intense work lasting roughly 30–75 minutes, when you do not want to or cannot swallow — in the heat, with a sensitive stomach, on a finishing stretch, during critical intervals.
- Protocol: a maltodextrin solution of around 6–6.5%, rinse for 5–10 seconds, then spit. Longer and sweeter adds nothing.
- What not to do: use it to replace real fuelling on long distances. In a marathon and beyond it is swallowed carbohydrate that decides the outcome, not a receptor signal. The calculator below will help you work out your fuel needs.
- Expectations: this is not “a minute faster”, it is fine-tuning. With an effect of g = 0.18 it is reasonable to expect fractions of a percent, not a breakthrough.
Limitations
The authors themselves rated the certainty of evidence as low for performance overall; it rose to “moderate” only for the five-second rinse variant. There are two reasons: inconsistency of results and detected publication bias (Egger's test, p = 0.02) — meaning null-result studies were probably published less often, and the true effect may be smaller still than calculated. Add a strong male skew in the sample (69 of 73 effect estimates) and inconsistent reporting of participants' nutritional status, and you get an honest picture: the technique works, but modestly and capriciously.
Key points
- Rinsing acts through oral receptors and the brain, not through absorption — the effect is instant and does not burden the gut.
- The overall effect is small: g = 0.18. The prediction interval includes negative values — it may not work for a given individual.
- Optimum: maltodextrin ~6–6.5%, 5–10 seconds. More concentration and more time add nothing; sucrose barely works.
- The “only works when fasted” rule rests on just four contradictory comparisons — it cannot be considered established.
- Certainty of evidence is low and publication bias is present — the real effect may be smaller.
- A tool for short intense efforts, not a substitute for fuelling in long races.
Sources: “Fed, not fasted: is carbohydrate mouth rinsing still ergogenic? A three-level meta-analysis”, Journal of the International Society of Sports Nutrition, 2025. https://doi.org/10.1080/15502783.2025.2579027; https://pmc.ncbi.nlm.nih.gov/articles/PMC12599167/