Skip to content

Caffeine mouth rinsing showed small exercise effects and uncertain cognitive effects

A meta-analysis found a very small exercise improvement, while overall cognitive results were inconclusive and changed when unusually extreme findings were removed.

By 100HP editorialAbstract-based explanation checked

Based on the published abstract. The full paper may contain additional methods, results and limitations.

The 30-second takeaway

Across the included studies, caffeine mouth rinsing was associated with a statistically significant but very small improvement in exercise performance. The main analysis did not find a statistically significant overall cognitive benefit. Cognitive results changed after unusually extreme findings were excluded, so conclusions about thinking performance were less stable than the exercise result.

Keep in mind

Evidence certainty was moderate for exercise and very low or low for cognition, with bias concerns common. This abstract-based summary cannot identify the comparison conditions, participant characteristics or follow-up periods because those details are absent.

How the evidence was tested

Researchers pooled results from exercise and cognitive tasks, then examined whether testing conditions and participant characteristics helped explain differences. They repeated analyses after removing unusually extreme results, called outliers, to test how much those results influenced conclusions.

Context changed the picture

Exercise findings were most consistent for aerobic endurance and when participants had eaten. After outlier removal, overall cognition and speed-related results became statistically significant, while accuracy remained variable. The suggested exposure-response pattern did not hold up after outlier removal.

CHECK THE ORIGINAL

The original publication

Revisiting the evidence on caffeine mouth rinse: effects on exercise and cognitive performance: a meta-analytic review.

Deng H, Fan X, Song T et al.
J Int Soc Sports Nutr · 2026

PubMed ID
41774525
Record checked

AI-assisted research and writing. This explains one selected publication; it is not a complete review of everything known. Our approach.