REF-748 · Burke, Dolan, Gonzalez, Mujika, Whitfield et al. (UCI Sports Nutrition Project), 2026 · International Journal of Sport Nutrition and Exercise Metabolism, ahead of print · Paper
The UCI position statement on nutrition for cycling: one governing-body consensus that puts numbers on energy availability, daily carbohydrate by training load, on-bike intake by race duration, recovery, hydration, bone and the six supplements with a case
And says plainly what it is not sure of
What it is
A summary compiled by an organising committee (Burke, Dolan, Gonzalez, Mujika, Whitfield) from 14 themed narrative reviews, each separately peer-reviewed in IJSNEM, then endorsed by the contributing authors and the UCI representative (Bigard). Ten themes: discipline demands, training adaptation, race preparation, on-bike nutrition, physique and energy, illness and injury, culinary support and environments, technology, supplements, age and sex.
Key Findings — the numbers
Energy availability: aim for a habitual 40–45 kcal·kg FFM⁻¹·day⁻¹ (lower end for riders with little activity outside training). Total intake may need to reach 7,000–9,000 kcal·day⁻¹ in stage racing. Short, planned periods of low EA are acceptable if recovery follows and habitual adequacy is maintained.
Daily carbohydrate, indexed to training load: light endurance 5–7 g·kg⁻¹ · track 3–8 · moderate endurance (<3 h·day⁻¹) 6–10 · intensified (>4 h·day⁻¹) >8 · extreme racing (>6-h races, stage events) 10–25 g·kg⁻¹. "Train-low" sessions may amplify adaptation but the evidence that this reaches performance "is currently limited"; they must not compromise key sessions or habitual EA.
Protein 1.6–2.1 g·kg⁻¹·day⁻¹, spread across the day, upper end during low-carbohydrate or energy-restricted periods; timing and distribution benefits are "equivocal" but practical for hitting totals. Fat ~20–35% of energy, inversely to carbohydrate on the day.
Race preparation: carbohydrate loading 10–12 g·kg⁻¹·day⁻¹ for 1–2 days before single-day events; pre-race meal 1–4 g·kg⁻¹ in the 1–4 h before; start euhydrated. A low-residue diet for 48–72 h drops body mass by about 0.74 ± 0.99% without fuel or fluid deficit — a tool before a mountain stage, not a habit.
On-bike carbohydrate, scaled to duration: 40–75 min — mouth rinsing (5–10 s contact); 1–2.5 h — 30–60 g·h⁻¹; >2.5 h — 90 g·h⁻¹; 120 g·h⁻¹ and above "are noted but require validation of benefits". Above 60 g·h⁻¹ use glucose + fructose products and train the gut.
Fluid: keep the net deficit <2% body mass, especially in heat; strategies range from ad libitum to a plan based on measured sweat rates. Deliberate dehydration for power-to-mass "may not actually" pay off once the fluid deficit costs performance.
Recovery between stages or races: carbohydrate 1–1.2 g·kg⁻¹·h⁻¹ starting soon after; protein 0.3–0.4+ g·kg⁻¹; after a >2% deficit, fluid at 125–150% of the loss over 4–6 h with sodium (~50–80 mmol·L⁻¹ or salted food).
Bone: road cyclists have lower bone mass than other athletes; calcium 800–1,200 mg·day⁻¹ with a possible benefit from pre-exercise calcium; 25(OH)D >50, ideally 75 nmol·L⁻¹; low CARBOHYDRATE availability, not only low energy, may drive the acute bone-marker changes seen in short-term deficits.
Iron: monitor periodically; 100–200 mg·day⁻¹ elemental iron when treating deficiency or supporting altitude blocks. Cold raises daily energy expenditure by ~10%; altitude raises glycogen use and fluid loss.
Supplements: only six carry a case — caffeine, creatine, sodium bicarbonate/citrate, β-alanine, nitrate, glycerol (AIS Group A). Exogenous ketones: not recommended in any discipline. Caffeine is the only one ticked for every discipline; glycerol only for road; creatine and nitrate are ticked for BMX, track and Esports and marked "more evidence needed" for MTB, cyclo-cross, road and gravel. Most of the evidence is from Tier 2–3 males.
Technology: power meters are the one tool with established nutritional utility (energy expenditure). Continuous glucose monitors lag blood glucose by 10–15 min, misread during high-intensity exercise and after meals, differ by placement, and are banned by the UCI in competition; "it is presently unclear whether these data are relevant".
Female cyclists: "current evidence does not support generalized recommendations for altering training or dietary intake solely based on menstrual cycle phase" — track symptoms, treat symptoms. Menstrual disruption is an early indicator of low EA.
Masters: anabolic resistance argues for the upper protein band; bone nutrients matter more; thirst and sweat capacity decline, so hydration in heat needs a plan rather than thirst.
Read the direction carefully
This is a consensus of what a governing body's authors agree is defensible in 2026, not a study, and its authority is in the ranges and in what it declines to endorse. Three refusals are the most useful lines in it: 120 g·h⁻¹ is observed practice awaiting validation; train-low has cellular evidence and limited performance evidence; menstrual-phase periodisation has no general support. A coach who quotes the ranges must quote the refusals with the same confidence.
The numbers are written for professionals and scale by load, not by ambition. The extreme-racing carbohydrate band and the 7,000–9,000 kcal days describe Grand Tours; an amateur on 8 h·week⁻¹ lives in the 5–7 g·kg⁻¹ band and the 30–60 g·h⁻¹ row.
Practical Application
Index carbohydrate to the day, not to the athlete's identity. A 90-min session is a 30–60 g·h⁻¹ day and a 5–7 g·kg⁻¹ day; a 4-h ride is a 90 g·h⁻¹ day and a >8 g·kg⁻¹ day. State the row you are using.
Energy availability is the floor under everything else. Before prescribing a physique change, check that habitual EA is near 40–45 kcal·kg FFM⁻¹; treat body mass "as a metric rather than a goal" with periodised ranges, and screen with the REDs CAT2 (REF-341) when training intensifies.
Above 60 g·h⁻¹, the prescription is a product and a practice, not a number: glucose + fructose, and a gut-training progression in training sessions that mirror the race (REF-008).
Recovery between two sessions or two race days is a dose: 1–1.2 g·kg⁻¹·h⁻¹ carbohydrate early, 0.3–0.4 g·kg⁻¹ protein, and 125–150% of a measured fluid loss with sodium — not "eat well tonight".
For a cyclist, bone is a nutrition problem as much as a loading problem: calcium, vitamin D status, and carbohydrate availability on long days; a masters rider gets the upper protein band on top.
Supplements: caffeine for everyone, ketones for no one, the rest by discipline — and every product through third-party batch testing, because the strict-liability rule does not care who made the label.
Do not fuel by a CGM. Its lag and its intensity error make it a research tool at best; the power meter is the instrument that already tells you the energy cost.
For women, work from a symptom log, not from a phase calendar (SYNTH-37); a lengthening or missing cycle during a hard block is a fuelling question first.
Limitations
A narrative consensus, not a systematic review; the strength of each recommendation is only as good as the companion review behind it, and the statement itself does not grade them.
Written for elite riders and professional teams with culinary nutrition support; translation to self-supported amateurs is named as an open research question, not answered.
The supplement evidence base is acknowledged to be mostly Tier 2–3 males; discipline-specific effects in Tier 4–5 and in women are largely untested.
Table 1 lists an off-road energy allowance of "+24 kcal·kg⁻¹·km⁻¹" for XCO/XCM; as printed that figure is implausible by two orders of magnitude and is not carried into this entry.
Ahead-of-print at the time of entry; page numbers may change.
How this entry was checked and changed
Content correction 2026-09-27
Cross-reference pointed at an entry whose content does not match the label (checked against the cited entry); repointed or relabelled.
Cross-reference description repeated a claim that the cited entry no longer makes (it was corrected against its abstract); description updated to the cited entry's current content.
Cross-reference descriptions checked against the cited entries' current content; wrong years, wrong entries and claims the cited entry does not make were corrected.