How much do elite ultrarunners eat and spend in a hot 100-miler?
One world-class runner: 16,104 kcal spent, ~6,400 eaten in race (86 g carbs/h), no distress reported, −4.3% mass. One case, not a target.
Mougin and colleagues measured Kilian Jornet through the 2025 Western States Endurance Run, a 161-km mountain race with 5,500 m of climbing and air temperatures up to 40 °C, which he finished third in 14:19:22, 8 min behind the winner and 10 min outside the course record.
Energy expenditure came from doubly labelled water drunk the evening before, with urine sampled before the start and 1 h 40 min after the finish. He spent 16,104 kcal, about 18.8 kcal a minute, the highest total the authors know of over such a duration in ultra-running, against 6,720 kcal eaten (a deficit of 9,384 kcal).
He ate a mixed fat-and-carbohydrate food early and moved to gels later: 86 g of carbohydrate an hour on average and 110 g/h over the last 5 h, with no gastrointestinal distress reported. He drank 12.5 L (0.87 L an hour) with 18.5 g of sodium and still finished 2.5 kg (4.3%) lighter, with concentrated urine.
His gut temperature averaged 37.1 °C, stayed above 38 °C for 204 min and peaked at 39.4 °C in the final kilometres, when he ran fastest, not when the air was hottest. Urine markers of kidney stress rose (NGAL 9.4 → 25.4 ng/mL, KIM-1 0.30 → 1.70 pg/mL) with some protein and blood in the urine.
A speed model fitted to his training GPS data gave a critical speed for every gradient. In the race he averaged 84.8% of it, held about 87.4% for the first 5 h 07 min, then lost 1.45 percentage points an hour to finish at 74.0% — a 15% decline.
Our reading: a well-measured reference for the top of the range — how much a world-class runner spends, eats and drinks in a hot 100-miler — and a demonstration of a field method for pacing against grade-specific critical speed. It is one athlete, self-selected intakes and no comparison, so it shows what is possible for him, not what anyone else should copy, and nothing here says what caused his result.
Protocol
Who: n = 1, a male professional mountain ultrarunner, 37 y, 1.72 m, 58 kg, BMI 19.6 kg/m², body surface area 1.66 m²; fat-free mass 43.3 kg (75%) and total body water 31.70 kg (55%) from the isotope dilution. His training logs are public.
Race: 2025 Western States Endurance Run, 161 km, 5,500 m of cumulative climbing, air temperature up to 40 °C (at 120 km, 11 h in), relative humidity at most 60%.
Heat preparation (self-reported): 12 exposures over the 3 weeks before the race (4 a week, 1.5-2.5 h each) in a makeshift chamber at 40-45 °C, aiming for a rectal temperature of 38.5-39.0 °C.
Energy expenditure: doubly labelled water (31.5 g H₂¹⁸O and 1.6 g ²H₂O) drunk at 19:30 the evening before; urine before the dose, 1 h 45 min before the start and 1 h 40 min after the finish; CO₂ production from the two elimination rates and expenditure by the modified Weir equation with an assumed respiratory quotient of 0.86 — an abbreviated two-point protocol the authors cite as validated for short field measurements.
Intake: food and fluid ad libitum; Table 1 ("Nutrition strategies", "Estimated intake") gives rounded amounts by phase, and the paper does not say how intake was recorded.
Core temperature: a telemetric pill swallowed 30 min before the start, sampled every minute; drinking artefacts smoothed with a maximum envelope.
Hydration and kidneys: body mass before and 5 min after the finish (before eating or drinking); urine osmolality and specific gravity; urinary NGAL and KIM-1 by ELISA; dipstick for albumin, protein, blood and ketones.
Heart rate and pacing: a COROS Apex watch and an optical arm band (1 Hz). A speed-gradient-duration model (a three-parameter critical-intensity model combined with a speed-grade polynomial) was fitted to the 99th-percentile mean speeds of 77 training sessions (12 June-27 August 2025) across gradients of −40% to +40% and durations of 10 s to 12 min.Race data were cut into 100-m segments, 39 outlier segments (2.4%, mainly aid-station stops) removed, and each segment's speed expressed as a percentage of the critical speed for its gradient.
Funding and interests: supported by the Western States Endurance Run Foundation; the pacing model was developed in a project funded by the French National Research Agency; the doubly labelled water analysis was paid for by the athlete, who is a co-author. Two authors are employed by Calorify, the company that ran the energy analysis; they state they did it without expecting a publication.
Key Results
| Measure | Value |
|---|---|
| Finish | 14:19:22, third; 8 min behind the winner, 2 min behind second, 10 min off the course record |
| Total energy expenditure (doubly labelled water) | 16,104 kcal (67.38 MJ); 13,492 kcal per 12 h; about 18.8 kcal/min |
| Energy intake | 6,720 kcal: 1,290 g carbohydrate, 190 g fat; deficit 9,384 kcal (intake about 40% of expenditure) |
| Carbohydrate rate | 86 g/h on average; hours 0-4 a 50/50 fat-carbohydrate mix (~250 g carbohydrate and ~110 g fat, about 62 g/h carbohydrate, our arithmetic); 80 g/h (+10 g/h fat) in hours 4-9; 110 g/h in hours 9-14.3 |
| Fluid and sodium | 12.5 L (0.87 L/h on average; 1 L/h with 1.5 g/L sodium after hour 4); 18.5 g sodium |
| Body mass | 58.0 → 55.5 kg (−2.5 kg, −4.3%) |
| Urine | specific gravity 1.005 → 1.015; osmolality 120 → 790 mOsm/kg H₂O |
| Gut temperature | mean 37.07 °C, peak 39.44 °C; above 38 °C for 204 min, above 38.5 °C for 168 min, above 39 °C for 63 min; 39.77 °C within 10 min of finishing |
| Kidney markers | NGAL 9.4 → 25.4 ng/mL; KIM-1 0.30 → 1.70 pg/mL; albumin 10 → 30 mg/L; blood 0 → 10 cells/µL; protein 0 → 20 g/L (as printed; 20 g/L would be far beyond the "moderate proteinuria" the authors describe, so the unit is probably a slip, our reading); ketones 0 → 0.5 mmol/L |
| Pacing vs grade-specific critical speed | mean 84.8%; about 87.4% held to 5 h 07 min, then −1.45 percentage points an hour; 74.0% at the end (−15%); 24 segments (1.4%) above 105% of critical speed, 15 (0.9%) without the final kilometre |
| Model fit | training profile r² = 0.939 (RMS error 1.06 km/h); race speed-grade r² = 0.723; pacing trend r² = 0.864 |
Comparisons the authors draw: 13,513 kcal per 12 h against 8,310 ± 1,520 kcal over ~12 h at the IRONMAN World Championship and 7,040 ± 1,360 kcal over ~12 h in recreational Western States runners; a 15% decline against a −32.0% ± 7.8% fall in flat-segment speed between the first and second halves for all UTMB finishers (a different measure).
Numbers that do not match (our check): the Discussion's 13,513 kcal per 12 h differs from Table 2's 13,492 (16,104 kcal over 14:19:22, i.e. 14.323 h, × 12 gives 13,492, matching Table 2; dividing by the rounded 14.3 h gives 13,514, the likely source of the Discussion's figure — our arithmetic); the Discussion's "theoretical expected range (~20 kcal·h⁻¹)" can only be per minute; Table 1 gives the last two phases' sodium once as "1.5 g·h⁻¹" and once as "1.5 g·L⁻¹" (at 1 L/h the amount is the same, and 1.5 g/L throughout reproduces the 18.5 g total, our arithmetic).Table 1's totals of 1,290 g carbohydrate and 6,720 kcal include the pre-race breakfast (60 g, 350 kcal); in-race intake was about 1,230 g and 6,370 kcal, and 1,230 g over 14.32 h is the reported 86 g/h (our arithmetic).The conclusion says intakes of ~90 g/h are "tolerable and sufficient"; the measured average was 86 g/h, and nothing was compared, so "sufficient" is the authors' reading.
No gastrointestinal distress was reported by the athlete; no symptom questionnaire is described.
Mechanisms
Measured: gut temperature peaked in the final kilometres, when he ran fastest (around 19:00, air about 35 °C), not at 11 h, when the air reached 40 °C.
Authors' interpretation: exercise intensity, more than air temperature, drives heat strain; his core stayed below 39.5 °C during the race because ultra running produces less metabolic heat and because of his acclimation and "precooling strategies"; the Results describe cooling from 39 km on (wet clothing, river immersion, ice slurry, menthol rinse, chilled gels). Nothing tested the cooling or the acclimation against their absence.
Authors' interpretation: his expenditure per hour exceeds amateurs' because he runs faster and walks less, and heat and dehydration raise the metabolic cost; the kidney markers and urine protein reflect transient stress, likely from muscle breakdown with heavy eccentric loading; the rises were lower than reported after a 120-km trail race. The post-race urine came 100 min after the finish, when drinking may have diluted it.
Authors' interpretation: he paced below critical speed almost throughout, which limited fatigue; the steady decline after 5 h reflects his critical speed itself falling with time, and he adjusted to stay below it.
Inference (our arithmetic): the paper's resting equation with his measured fat-free mass gives about 1,500 kcal a day at rest, so the race ran at roughly 18 times his resting rate, and the whole race day at roughly 11 times his daily resting expenditure — at or slightly above the ~10× that the metabolic-ceiling curve puts at one day (REF-959, via SYNTH-24; his resting rate is predicted by equation, not measured).The 2.5× figure is the curve's long-run asymptote, reached at about 28 weeks, not a limit on a single day.
Practical Application
How much do elite ultrarunners eat and spend in a hot 100-miler? One world-class runner: 16,104 kcal spent, ~6,400 eaten in race (86 g carbs/h), no distress reported, −4.3% mass. One case, not a target.
Fuelling (the base's reading): 86 g/h over 14 h, rising to 110 g/h, was possible for a gut-trained world-class runner in heat. It is the top of the range, not a starting rate; the base's start for a long run on foot stays lower and is built by rehearsal (SYNTH-02).
Expect a large deficit at any intake: even at 86 g/h he ate back about 40% of what he spent. Eating to match expenditure is not the aim in a 100-miler; eating what the gut tolerates is (inference).
Drinking: 0.87 L/h with about 1.5 g of sodium per litre still left him 4.3% lighter in up to 40 °C, above the ~2% the base treats as the heat limit (SYNTH-68). The authors wrote that higher intakes "would be required" but may not be practical; part of the mass lost over 14 h is fuel burned, not water (inference). One case cannot say what loss is harmful or whether drinking more would have helped.
Heat: he acclimated for 3 weeks (12 sessions at 40-45 °C, aiming for 38.5-39.0 °C core) and cooled from early in the race; his core peaked when he pushed at the end, not when it was hottest. The protocol is his, not a tested one; the base's acclimation rules are in SYNTH-04.
Pacing (method): a critical speed for every gradient can be fitted from training GPS data and the race read against it; he spent almost the whole race below it and slowed steadily after about 5 h (SYNTH-32). For athletes without such a model this is a description, not a tool yet.
Limitations
One athlete, one race. Every number is his; nothing separates the effect of his fuelling, cooling or acclimation from his talent, and the authors' "tolerable and sufficient" for ~90 g/h rests on one person who reported no distress, with no symptom questionnaire described.
Conflicts. The athlete is a co-author and paid for the doubly labelled water analysis; two authors work for the company that ran it.
Method limits. The abbreviated two-point doubly labelled water protocol assumes a respiratory quotient of 0.86, and its sampling window ran from 03:15 to 21:00 (about 17 h 45 min, 3.4 h longer than the race); the paper does not say how the 16,104 kcal was assigned to the 14.3 h of racing, so the per-minute and per-12-h rates may be slightly high (inference); intake is given as estimated amounts by phase, with no recording method described; the heat protocol is self-reported; the post-race urine came 100 min after the finish; the heart rate was from an optical arm band; critical speed is the asymptote of a model fitted to efforts of 10 s to 12 min, and the training window (12 June-27 August 2025) brackets the late-June race, so most of the profile comes from training after it (inference from the race date).
Reporting. 13,513 vs 13,492 kcal per 12 h, "~20 kcal·h⁻¹" for a per-minute figure, and sodium written per hour and per litre for the same phases (see Key Results).
Comparisons across measures. The −32% UTMB figure is flat-segment speed between race halves for all finishers, not the same normalised measure as his −15%.
How this entry was checked and changed
Why this level
single case study (rubric), read from the full text (open access, CC BY 4.0). Measurements are careful, but one athlete in one race shows what is possible for him, not what is typical or what caused it; the athlete is a co-author and paid for the energy analysis
Full text (checked 2026-10-26)
The athlete completed the 2025 WSER 100 in 14:19:22, finishing third; 8 min behind the winner and 10 min behind the course record (full text)
Total energy expenditure was 16,104 kcal; energy intake totaled 6,720 kcal (86 g carbohydrate per hour), an energy deficit of 9,384 kcal (full text)
Average carbohydrate consumption was 86 g per hour across the race, peaking at 110 g per hour in the last 5 h; 1,290 g of carbohydrate and 190 g of fat (full text)
The athlete consumed 12.5 L of fluids (0.87 L per hour; 18.5 g sodium) and lost 4.3% body mass, a 2.5 kg reduction (full text)
Gastrointestinal temperature averaged 37.07 °C and peaked at 39.44 °C; above 38 °C for 204 min (full text)
NGAL increased from 9.4 to 25.4 ng/mL and KIM-1 from 0.30 to 1.70 pg/mL (full text)
Mean normalized speed was 84.8% of predicted critical speed; about 87.4% for around 5 h, then a decline of 1.45% every hour to 74.0% (full text)
H.M. and E.S. are employed by Calorify; all analyses were funded by K.J.B., who was the study participant (full text)