All articles
6 min readrunning / endurance

Most runners race on 22 g of carbs an hour: fix your fueling

I run about 5,000 km a year and race 100 km. The document I spend the most time on before a long race is the eating plan. Two numbers frame the whole debate: recreational marathon runners swallow around 22 g of carbohydrate an hour, and the fueling advice circulating in trail running now opens at 120 g. Seven studies put the useful range somewhere between them.

Runners eat less than they planned and remember eating more.

Asli Devrim-Lanpir and colleagues tracked endurance athletes through a real race day and compared the plan with the stomach. Actual intake came in at 31.7 ± 23.5 g/h against a planned 38.0 ± 27.3 g/h 1. Marathon runners sat lowest, at 21.7 g/h eaten for 25.9 g/h planned, while cyclists reached 49 g/h. Gels had the highest leftover rate: carried for four hours, never opened. Athletes also credited themselves with more than they had taken.

Field data from a decade earlier points the same way. Beate Pfeiffer, Trent Stellingwerff, Asker Jeukendrup and co-authors recorded intake in 221 athletes across two Ironman races, a half Ironman, a marathon, a 100/150 km cycling race and a professional stage race. Mean carbohydrate intake in the three triathlons was 62 ± 26, 71 ± 25 and 65 ± 25 g/h, with individuals spread from 6 to 136 g/h 2. In both Ironman races, a higher intake went with a faster finishing time. It also went with nausea and flatulence, and 31% of Ironman athletes reported serious gastrointestinal symptoms.

The ceiling sits in your intestine.

Glucose crosses the intestinal wall through the SGLT1 transporter, which saturates around 60 g/h. Roy Jentjens and his group in Birmingham tested what a second door does to that limit. Cyclists rode for 120 min while drinking glucose alone or a glucose and fructose mixture, fructose being absorbed by GLUT5. Peak oxidation of the ingested carbohydrate reached 1.26 ± 0.07 g/min on the mixture, against 0.80 and 0.83 g/min on glucose alone, roughly 55% higher 3. That mechanism is why sports drinks and gels now print a glucose to fructose ratio on the label.

Performance stops improving around 78 g/h.

Smith and co-authors mapped the dose-response curve. Fifty-one cyclists and triathletes at four research sites completed four trials each: two hours at constant load, then a simulated 20 km time trial. Twelve drinks delivered 10 to 120 g/h in a 2:1 glucose to fructose ratio. The model predicted gains up to about 78 g/h, with diminishing returns and projected losses above that, and a flat response between roughly 30 and 90 g/h 4.

Daniel Plews and colleagues reviewed the case for ultra-high intakes for Sports Medicine this year. Their verdict: the performance benefit is well established between 60 and 90 g/h, while above that band the evidence stays limited and inconsistent, and the standard explanations (glycogen sparing, blood glucose maintenance, maximum exogenous oxidation) fail to account for the claimed gains. Feed a group 120 g/h under standardized conditions and measured oxidation spreads between about 1.3 and 1.9 g/min from one athlete to the next 5.

The 120 g/h case rests on 20 elite runners with trained guts.

Aitor Viribay and co-authors randomized 20 male elite runners in a mountain marathon to 120, 90 or 60 g/h. Everyone in the trial had completed nutritional and gut training first. The 120 g/h group finished with lower creatine kinase, lactate dehydrogenase and aspartate aminotransferase, three markers of muscle damage 6. A 2021 systematic review of eight studies on one-day ultra-trail races, by Soledad Arribalzaga and colleagues, found the same direction of effect, plus two figures worth keeping: 65% to 82% of runners reported gastrointestinal symptoms, and most of them never reached the 90 g/h recommendation 7.

Twenty gut-trained elites make a thin foundation for a number now printed on packaging. Treat 120 g/h as a ceiling that a minority of athletes reach after months of preparation.

Your gut adapts in two weeks.

Ricardo Costa and his Monash group put 25 endurance runners through a two-hour gut challenge, eating 30 g of carbohydrate every 20 minutes while running, followed by a one-hour distance test. Two weeks of repeated gut challenges cut gastrointestinal symptoms by 60% in the group using carbohydrate gels (p = 0.008) and 63% in the group using carbohydrate food (p = 0.046), both larger reductions than placebo 8. The stomach and the intestine respond to training on a fortnightly timescale, which is faster than your aerobic system.

The protocol.

  1. Set the rate from the duration. Under 60 min: water. 60 to 150 min: 30 to 60 g/h. Beyond 150 min: 60 to 90 g/h. Aim at 78 g/h if you want one number to plan around.
  2. Buy the ratio, not the brand. Check that the label shows glucose (or maltodextrin) with fructose, in a 2:1 or 1:0.8 ratio. Single-source glucose products cap you near 60 g/h whatever the plan says.
  3. Convert grams into objects, then into alarms. A 25 g gel every 20 minutes gives 75 g/h. Set a repeating 20-minute alarm on your watch and eat at the beep, before thirst or hunger arrives.
  4. Count the wrappers after every long run. Compare what you carried with what came home. This is the single check that separates your plan from the 22 g/h average.
  5. Train the gut for two weeks before you raise the rate. On your two longest sessions each week, eat at full race rate for the whole session. Expect the first one to feel heavy.
  6. Earn the right to go above 90 g/h. Two long runs at 90 g/h without symptoms, then test 100 to 110 in training. Never in a race.
  7. When symptoms start, drop to the rate you have rehearsed. Cutting to zero costs you the second half of the race.

On your next long run, carry one gel more than the plan requires, set the alarm at 20 minutes, and count what comes back in your pocket.

Sources.

  1. Devrim-Lanpir, A. et al. (2025). Under Consumed and Overestimated: Discrepancies in Race-Day Carbohydrate Intake Among Endurance Athletes. European Journal of Sport Science. doi:10.1002/ejsc.70055
  2. Pfeiffer, B., Stellingwerff, T., Hodgson, A.B., Randell, R., Pöttgen, K., Res, P., Jeukendrup, A.E. (2012). Nutritional intake and gastrointestinal problems during competitive endurance events. Medicine & Science in Sports & Exercise, 44(2), 344-351. doi:10.1249/MSS.0b013e31822dc809
  3. Jentjens, R.L.P.G., Moseley, L., Waring, R.H., Harding, L.K., Jeukendrup, A.E. (2004). Oxidation of combined ingestion of glucose and fructose during exercise. Journal of Applied Physiology, 96(4), 1277-1284. doi:10.1152/japplphysiol.00974.2003
  4. Smith, J.W., Pascoe, D.D., Passe, D.H., Ruby, B.C., Stewart, L.K., Baker, L.B., Zachwieja, J.J. (2013). Curvilinear dose-response relationship of carbohydrate (0-120 g/h) and performance. Medicine & Science in Sports & Exercise, 45(2), 336-341. doi:10.1249/MSS.0b013e31827205d1
  5. Plews, D.J., Booth, P.D., Krieger, T., Maunder, E. (2026). Fuelled or Fooled? Examining the Evidence and Mechanisms Behind Ultra-High Carbohydrate Intake in Endurance Athletes. Sports Medicine. doi:10.1007/s40279-026-02462-z
  6. Viribay, A., Arribalzaga, S., Mielgo-Ayuso, J., Castañeda-Babarro, A., Seco-Calvo, J., Urdampilleta, A. (2020). Effects of 120 g/h of Carbohydrates Intake during a Mountain Marathon on Exercise-Induced Muscle Damage in Elite Runners. Nutrients, 12(5), 1367. doi:10.3390/nu12051367
  7. Arribalzaga, S., Viribay, A., Calleja-González, J., Fernández-Lázaro, D., Castañeda-Babarro, A., Mielgo-Ayuso, J. (2021). Relationship of Carbohydrate Intake during a Single-Stage One-Day Ultra-Trail Race with Fatigue Outcomes and Gastrointestinal Problems: A Systematic Review. International Journal of Environmental Research and Public Health, 18(11), 5737. doi:10.3390/ijerph18115737
  8. Costa, R.J.S., Miall, A., Khoo, A., Rauch, C., Snipe, R., Camões-Costa, V., Gibson, P. (2017). Gut-training: the impact of two weeks repetitive gut-challenge during exercise on gastrointestinal status, glucose availability, fuel kinetics, and running performance. Applied Physiology, Nutrition, and Metabolism, 42(5), 547-557. doi:10.1139/apnm-2016-0453

Keep reading