Does more carbohydrate always improve cycling? An experiment shows why context matters

If an amount of carbohydrates can help sustain the exercise, it seems intuitive that increasing the dose will help even more. The problem is that a plausible relationship does not always grow linearly.

An experiment published in 2026 investigated this question by controlling the intensity of the exercise and comparing different quantities in the same participants. The conclusion requires more care than a “more is better” rule or “high doses do not work”.

How the question was tested

Twelve men with recreational training completed nine conditions: three cycling intensities combined with placebo, 40 or 90 g of carbohydrates per hour. The previous stretches lasted from 60 to 120 minutes and were followed by a time trial of 4 km.

After 90 minutes at the bottom of the heavy intensity domain, 40 g/h improved time compared to placebo: 8.3 versus 8.6 minutes. The dose of 90 g/h did not present this benefit. In other conditions, there were no significant differences. The research did not demonstrate that 40 g/h were universally superior to 90 g/h, nor did it evaluate a competition of many hours.

Why was the intensity a central part of the design?

Prescribing the same percentage of a maximum value does not guarantee that two people are in the same physiological domain. The researchers sought to better define the position of efforts in relation to individual thresholds.

For the reader, the lesson is that “moderate cycling training” can be an inaccurate description when comparing studies. A result depends on what effort was made, how long it lasted and how performance was measured afterwards.

It is also important to note that the previous feeding was standardized. The experiment does not represent all possible combinations of fasting, low previous consumption and different fuel stocks.

A different result does not automatically identify the mechanism

The authors discuss hypotheses to explain the absence of additional benefit, but do not demonstrate a single cause. In particular, gastrointestinal symptoms did not differ between conditions. Therefore, it would be incorrect to turn the study into evidence that the higher dose damaged the intestine and therefore worsened performance.

This distinction between finding and explanation is useful in any scientific reading. Note that two conditions had different results does not inform, alone, which stage of the process produced this difference.

Editorial example: two sessions, two questions

A cyclist can do a controlled session of 90 minutes during the week and a long cycling preparation training over the weekend. Evaluating the same strategy in both situations allows you to ask different questions. Is it executable? Is it well tolerated? Does it help complete the purpose of the session?

The experiment does not authorize filling in the response of long cycling training with the result of the short test. Nor does it allow discarding an individual strategy just because it was not superior in a small group of recreational men.

A more useful way to take advantage of the research is to bring the nutritionist the context of the task, rather than asking for a dose based on platoon trends. The question comes to include duration, intensity, previous feeding, tolerance and objective.

The research questions the automatic dose escalation. It does not take away the importance of carbohydrate and does not define a single amount for every practitioner. Its value is to show how a nutritional question needs to be tested under clearly described conditions.

Scientific reference

Recommended but Not High Carbohydrate Dose Improves Performance After Heavy- but Not Moderate-Intensity Exercise in Recreationally Trained Participants. Scandinavian Journal of Medicine & Science in Sports, 2026. DOI: 10.1111/sms.70288.

Editorial text in our own words. Examples identified as editorials are illustrative and do not represent additional participants or interventions of the research.

Cada um no seu ritmo. Juntos no caminho.