SYNTH-02 · Guide
Carbohydrate per hour
Pick the row by the session's duration and the column by the athlete's level, earn the number in training over weeks, and treat 120 g/h as a proposed ceiling for trained athletes, never as everyone's default
The question it answers
Default (what to prescribe first)
1
Duration picks the row, level picks the column, and the athlete's own row is the one spoken. The UCI statement scales on-bike intake to duration: 1-2.5 h 30-60 g/h, >2.5 h 90 g/h, and 120 g/h and above "are noted but require validation of benefits" (REF-748, A, written for elite riders). Not from a study; the base's own practice: an amateur's 3-6 h event is planned at 80-100 g/h.
2
Above ~60 g/h use glucose + fructose and train the gut (REF-748).
3
Earn the number in training. The gut is trainable, but the method and timeline are not established (REF-008, B+, narrative review). Not from a study; the base's own practice: start from today's intake and add 15-20 g/h a week.
4
Start early, feed often. Not from a study; the base's own practice: first intake by minute 15, then 20-30 g every 15-20 min rather than one large dose an hour; no cited entry tests the interval.
5
Drink to thirst — the whole decision is in SYNTH-68.
Decision — the matrix (duration × level)
Levels are the base's own bands: amateur VO2max 45-55, sub-elite 55-65, elite ≥65; no study compared intakes by VO2max. Inference: with no known level, use the amateur column.
| Duration | Amateur | Sub-elite | Elite | Source and how sure |
|---|---|---|---|---|
| <90 min | From ~1 h: 30-60 g/h; shorter: small amounts or a mouth rinse are enough | same | same | Consensus: 1-2.5 h 30-60 g/h, and for 40-75 min a mouth rinse (REF-748, A); ~1 h of sustained high intensity needs only small amounts or a rinse (REF-066, B+); in 12 male amateur cyclists a maltodextrin or fructose rinse gave no more power in a 15-min time trial than placebo or no rinse (REF-957, B) — the rinse is optional (the base's own reading; efforts near 1 h were not tested) |
| 90 min-3 h | 60-90 g/h | 80-100 g/h | 90-120 g/h | Amateur cell is the base's own practice, set at and above the top of consensus, which is 30-60 g/h for longer events, up to 90 g/h beyond 2.5 h (REF-066; REF-748); 60 g/h negated the 9% fall in critical power after 2 h of heavy cycling (REF-145, B+, n=16); after 3 h of moderate riding neither 60 nor 120 g/h did (critical power −7.4% at 60 g/h and −4.2% at 120 g/h vs −14.7% on water, REF-1122, B, n=16). Sub-elite and elite cells are the base's own practice, scaled toward the proposed 120 g/h ceiling (REF-3490, B+) |
| 3-6 h (gran fondo, gravel, marathon — runners: inference, expect the low end or less, see the gut ladder) | 80-100 g/h | 100-120 g/h | 110-120 g/h | Consensus is 90 g/h beyond 2.5 h (REF-748) and ~90 g/h for runs ≥3 h (REF-663, B+); all three cells are the base's own practice around that anchor and the proposed ceiling (REF-3490). 120 g/h was measured in elite mountain runners (REF-331, REF-332, B+) and 9 trained male cyclists (REF-333, B); 100 g/h in 12 elite cyclists over 4 h (REF-144, B) |
| Multi-day stage racing (4-6 h stages) | 90-110 g/h | 100-120 g/h | 100-120 g/h | The base's own practice; ≥100 g/h is reported professional-peloton practice whose plausible case is recovery across days, not a single-day gain (REF-334, C+). Only the 120 g/h arm avoided a significant 24-h drop in jump and half-squat performance after a trail marathon (REF-332); what to eat between the days: SYNTH-85 |
| >6 h single effort (ultra) | Start lower, set by rehearsal: on the bike begin at 60-80 g/h; on foot begin nearer the ultra-running stand's 30-50 g/h and add only what rehearsal shows the gut holds; the 120 g/h ceiling does not apply | same | same | The base's own practice. Reported intakes in races of ~12-24 h are 30-66 g/h, elite case studies 80-100 g/h (REF-3490, full text) (corrected 2026-10-12); one world-class mountain runner averaged 86 g/h over 14 h 19 min of a hot 100-mile race and 110 g/h over the last 5 h, with no distress reported (REF-1194, C+, case study — what one athlete took, not a starting rate); the ultra-running stand gives 30-50 g/h (REF-528, A, running); 14 riders over 1,230 km averaged 57.1 g/h, falling after half-way (REF-799, B); ~90 g/h tolerability in ultra competition is "questionable" (REF-663) |
| Any row, hot day | Inference from one trial: keep the row's rate if rehearsed in heat; do not raise it; expect less of it oxidised | same | same | One trial, 32 °C vs 19 °C (REF-461, B, 10 trained men) — see "In the heat" |
Reading the matrix
Say the athlete's own row, not the elite one. REF-331: 20 gut-trained elite male mountain runners, 120 g/h gave lower CK, LDH, GOT and internal load than 60 or 90 g/h; race time not reported. REF-332: 26 elite trail runners across three arms. REF-333: 120 g/h tolerated with minimal GI distress after a day at 8 g/kg carbohydrate and a 2 g/kg pre-exercise meal. A recreational athlete handed 110-120 g/h has been given an elite number.
Above 120 g/h is reported behaviour, not a guideline: 120-200 g/h is "not yet substantiated by current scientific research" (REF-3490). In 8 trained men riding 3 h, 150 g/h raised exogenous oxidation over 120 g/h (1.62 vs 1.48 g/min) but not whole-body carbohydrate oxidation (2.6 g/min both) and gave no longer capacity test than 90 or 120 g/h; the modelled best dose was 107 g/h (95% CI 92-122) (REF-1103, B).
100-120 g/h is "oxidised more", not "proven faster". The ceiling is argued from oxidation, not finishing time (REF-3490); the few comparisons of ≥100 with 60-90 g/h show no clear single-ride benefit (REF-334).One later crossover found a fatigued-capacity benefit: after 3 h at 95% of the gas exchange threshold, 3-min-test critical power was 266 W at 120 g/h vs 257 W at 60 g/h (fresh 277 W), W′ fell with no difference between doses (REF-1122, B, n=16, drink, water control, participants not reported blinded to dose); it did not test 90 g/h and measured no finishing time.
Where athletes actually are. In 221 athletes surveyed after races, intake ranged 6-136 g/h; means were 62-71 g/h in Ironman, 53 g/h in a 100/150-km cycling race and 35 g/h in the marathon (REF-787, B+, self-report).
Composition and form
Ratio: at or below ~60 g/h a single source is enough; above it, a fructose-to-glucose ratio of 0.6-1.0 (REF-3490, full text). REF-333 used 1:0.8 maltodextrin/glucose:fructose at 120 g/h without comparing ratios.Why a blend: glucose alone (drunk at 1.2 g/min, half the blend's dose) peaked at 1.06 g/min of exogenous oxidation against 1.75 g/min for a 1:1 blend drunk at 2.4 g/min (REF-527, B, 8 trained male cyclists, oxidation only); that fructose uses a second transporter is inference, not tested there.
Form is preference: at 120 g/h, drink, gel, chew or a mix gave the same peak exogenous oxidation (1.56-1.66 g/min) and subsequent exercise capacity (REF-333).
How to progress / periodise — the gut ladder
The ladder. Not from a study; the base's own practice: week 1 at the current intake (often 40-50 g/h); weeks 2-3 add 15-20 g/h each; week 4 hold the target rate for 1 h; week 5 for 2 h; week 6 over the full race duration as a rehearsal. The ladder ends at the athlete's own cell, not at 120 g/h.
Why weeks. Gut training at 30-90 g/h reduced discomfort in the reviewed interventions (REF-3490, full text). Inference: one session is unlikely to do it.
GI history lengthens the ramp (inference). Symptom scores tracked a history of GI distress (r = 0.37 and 0.51) (REF-787). Significant GI disturbance appears from ≥2 h at 60% VO2max irrespective of fitness, and running and heat worsen it (REF-525, A-, systematic review). Inference: runners should expect a lower tolerable rate than riders.
Race morning. Pre-race meal 1-4 g/kg in the 1-4 h before the start (REF-748); a ~2 g/kg breakfast with about a third from fructose extended time to exhaustion by ~7 min in 8 trained men (REF-065, B, a tweak). Loading and the full day: SYNTH-69, SYNTH-64. Not from a study; the base's own practice: a familiar low-fibre meal and a gel 15 min before a hard start.
In the heat
Measured, once. In 10 trained male cyclists/triathletes riding 2 h at ~50% Wpeak with fan airflow of ~29-35 km/h and 72 g/h of glucose, 32 °C versus 19 °C cut exogenous glucose oxidation by 14-15% and made a ~15-min time trial 15% slower, with no difference in GI comfort or I-FABP (REF-461, B). It tested no temperature in between, so the base gives no per-degree adjustment.
What to do. Inference from REF-461: because gut comfort and integrity did not differ, cutting intake "to protect the gut" is not supported — keep the row's rate, do not raise it, and pace by RPE and heart rate, not cool-weather power.Not from a study; the base's own practice: use a glucose + fructose blend (its advantage is temperate data, REF-527; untested in heat), take more of it as liquid, and rehearse the race fuel in the heat — REF-008 says to rehearse in the conditions the race will bring.
When NOT to use it / red flags
Fluid is not set here. Drink to thirst by default and never more than you lose — the international consensus puts every known death from exercise-associated hyponatraemia on drinking beyond losses (REF-776, A-); the decision, and sodium, are in SYNTH-68. Not from a study; the base's own practice, as a planning range only: 500-800 mL/h, up to ~1000 mL/h in heat above 28 °C — a ceiling capped at the athlete's weighed sweat loss, never a target to hit.
Do not prescribe 120 g/h for a day longer than 6 h; that goes beyond what has been measured (REF-3490, full text).
Never a new rate or product on race day (REF-008: practise race fuelling in training). Not from a study; the base's own practice: no concentrated gels without fluid.
What the evidence does and does not show
Does not show: that 120 g/h is faster than 90 g/h (REF-3490, REF-334; capacity after 3 h did not differ between 90, 120 and 150 g/h in REF-1103; REF-1122 found 120 > 60 g/h for critical power after 3 h but did not test 90 g/h); any comparison of rates across athlete levels (one trial tested 60 vs 120 g/h in riders inside the amateur band, REF-1122); any rate tested beyond ~6 h; anything in the heat beyond one glucose-only contrast (REF-461).The mechanism is itself disputed: one argued review holds that carbohydrate works by preventing hypoglycaemia and gives no dose (REF-476, B+).Pooled crossovers show carbohydrate slightly spares muscle glycogen (~24 mmol/kg dry weight over ~110 min, 95% CI 4-45) and that ingestion rate did not significantly moderate the sparing (REF-1123, B+; ≥100 g/h rarely tested), so a higher row is not justified by muscle sparing.
Evidence gaps
No trial compares grams per hour by athlete level or fitness.
No controlled comparison of carbohydrate rates in events longer than ~6 h.
Glucose + fructose at 90-120 g/h in the heat is untested; so are heat acclimation × exogenous oxidation and any temperature between 19 and 32 °C.
The timeline of gut adaptation, the feeding interval and ratio comparisons at 120 g/h are in no cited entry.
How this entry was checked and changed
Content correction 2026-09-27
Matrix presented as sourced -> labelled "Not from a study; the base's own practice" with its three anchors named; elite 3-6 h cell "110-140" and >6 h row "90-110 / 110-130 / 120-160" -> capped at the 120 g/h ceiling of REF-3490; ultras split out (field 30-66 g/h, elite case studies 80-100 g/h per the REF-3490 entry; ~90 g/h tolerability questionable per REF-663); stage racing ≥100 g/h stated as reported practice (REF-334).
"Gut earns the number over 4-6 weeks" -> REF-008's corrected entry says the adaptation timeline is not established; the six-week ladder, feeding interval, fluid/sodium targets, race-morning protocol and pitfalls labelled base practice.
"Sweat sodium varies 200-1800 mg/L (REF-221)" -> abstract carries 1.28 ± 0.57 L/h and 51.7 ± 27.8 mmol/h for endurance athletes in 1303 athletes; the 20-80 mmol/L concentration range is entry-only, restated with arithmetic.
Full-base pass (later 2026-09-27): "the REF-221 entry gives ~40 mmol/L sweat sodium, 20-80 mmol/L range (≈460-1840 mg/L)" -> that concentration range was removed from REF-221 (not in its abstract); replaced by the abstract's sodium-loss rate, 51.7 ± 27.8 mmol/h (≈1190 ± 640 mg/h by arithmetic). "140-160 g/h peloton reports carried in the REF-334 entry" -> those figures are no longer in REF-334; wording updated.
Residual sync (REF-3490 re-checked): "revised ceiling"/"the guideline revision" -> "proposed" revision (perspective review, B+); ">80% Tier 1-2 participants" labelled full text; "adaptation is transporter-side and needs repetition over weeks" -> the authors' reading (transporter adaptation) plus the weeks-of-repetition point labelled inference.
2026-09-30: the fluid line gave 500-800 ml/h (up to ~1000 in heat) with no cap at measured loss and no mention of the hyponatraemia consensus; now leads with thirst and REF-776 and points to SYNTH-68, the range kept as a capped planning ceiling.
2026-10-01: rewritten in the decision-framework format; default and matrix cells unchanged; each row now carries its own source and practice label (REF-748, REF-066 under the consensus cells).Fluid and sodium -> SYNTH-68 (500-800 mL/h kept only as a capped ceiling; the 300-700 mg/L line and REF-221 data dropped); race morning "2-4 g/kg at T-3 h" -> 1-4 g/kg in the 1-4 h before (REF-748, as SYNTH-69). Absorbed from SYNTH-14: the hot-day row (REF-461); its temperature-band percentages and programmed 600-1000 mL/h drinking are not carried over.
Revision 2026-10-01: folded REF-957 (Mendonça 2026, B, double-blind crossover, 12 male amateur cyclists) into the <90 min row's source cell: maltodextrin and fructose rinses gave no gain over placebo or no rinse in a 15-min time trial. The matrix cells are unchanged; the rinse stays an option. Level unchanged (B).
Verification 2026-10-01: all 22 cited entries opened in the snapshot (REF-748, 066, 957, 145, 3490, 663, 331, 332, 333, 144, 334, 528, 799, 461, 008, 787, 527, 525, 065, 776, 181, 476).Every number quoted with a refId matches its entry (30-60 / 90 / 120 'require validation'; 9% CP; 30-66 and 80-100 g/h past 6 h [superseded 2026-10-12: 30-66 g/h is for races of ~12-24 h]; 30-50 g/h; 57.1 g/h; 6-136, 62-71, 53, 35 g/h; r = 0.37/0.51 and -0.55/-0.48; 1.06 vs 1.75 g/min; 1.56-1.66 g/min; 14-15% and 15%; ~7 min; ~10% after 4 h). No NUMBER or WRONG finding.
Verification 2026-10-01: matrix cell audit. Sourced: <90 min row for sessions of ~1 h and longer (REF-748/REF-066). Not sourced, and above the cited consensus: amateur 90 min-3 h (60-90 vs 30-60 g/h up to 2.5 h). Labelled practice and correctly so: sub-elite and elite 90 min-3 h, all three 3-6 h cells, all multi-day cells, the >6 h start. No study in the base compares rates by athlete level; the SYNTH says so.
Verification 2026-10-01: the >6 h start of 60-80 g/h is labelled practice but sits above the A-level ISSN stand for ultra running (REF-528: 30-50 g/h) and at the top of observed field intakes (REF-3490: 30-66 g/h); the cell is sport-blind. SYNTH-65 handles it ('fall back toward the stand's 30-50 g/h'); no pair proposed, flagged for the owner.
Verification 2026-10-01: downstream frameworks checked for consistency: SYNTH-50, SYNTH-58, SYNTH-64, SYNTH-65 repeat the matrix numbers unchanged; SYNTH-52 and SYNTH-64 already label the cells 'base practice'. If the <90 min, multi-day elite or marathon-row pairs are applied, SYNTH-50/58/64 copies need the same wording.
Verification 2026-10-01: REF-3490's 0.6-1.0 ratio, >6 h intakes and gut-training lines are 'full text, checked 2026-09-27' in the entry and are cited as full text in the SYNTH: OK. No cited entry is NOT IN SEARCH / UNVERIFIED / without DOI.
2026-10-01: the >6 h cell did not tell running from cycling although its own source column does (ultra-running stand 30-50 g/h); the starting rate is now split by sport, no new number.
Revision 2026-10-09: folded REF-1103 (Martyn 2026, B, randomised crossover, 8 trained men, 3 h): 150 g/h did not raise whole-body carbohydrate oxidation over 120 g/h, capacity was the same at 90/120/150 g/h, modelled best dose 107 g/h. Matrix cells, default and level unchanged.
Revision 2026-10-09: cross-references to SYNTH-85 (fuelling when the next hard session is close) added to the multi-day row and the cross-references. Matrix and level unchanged.
Revision 2026-10-13: folded REF-1122 (Norte 2026, B, randomised crossover, 16 trained cyclists/triathletes, 3 h at 95% GET; full text): critical power fell 14.7% on water, 7.4% at 60 g/h, 4.2% at 120 g/h; W′ fell, unaffected by dose; 90 g/h not tested; participants not reported blinded to dose.Folded REF-1123 (Rothschild 2026, B+, meta-analysis of 31 crossovers; full text): small muscle glycogen sparing, ingestion rate not a significant moderator. Tier and Level parentheticals updated; matrix cells, default and level unchanged.
Revision 2026-10-26: REF-1194 (Mougin 2026, C+, one world-class mountain runner at Western States: 86 g/h average, 110 g/h late, no reported distress) added to the >6 h row's evidence as the top of the observed range; the row's starting rates are unchanged; level unchanged.