Skip to content

Fueling, carbohydrate availability, and energy availability

Last checked: 2026-10-02

This is a living brief for a trained amateur cyclist training about 7–12 hours per week, with goals of VO₂max, FTP/threshold, and general fitness. It covers questions 13 and 14 — fueling and carbohydrate availability, and recovery nutrition and energy availability. It is not a medical prescription and not a copy of a professional 20–30 hour week.

No transcript was supplied. The prompt asks whether ordinary sucrose or sugar water can perform like commercial carbohydrate products. That is a hypothesis. It is not treated as a confirmed quotation from a named speaker or a named trial. Papers below are candidate evidence for the metabolic half of that question. None is a head-to-head of table sugar versus a named gel on gut comfort, sodium, fluid, and teeth.

Navigation: platform overview · terminology · bibliography

Reading order: Previous — threshold, durability and long rides · Next — sleep, heat and iron.

Question navigation: 13 Fueling · 14 Recovery and energy availability · Conclusions · 7–12 h versus 20–30 h · Open questions · References

How grades are used

Each conclusion is graded on its own, not the whole topic. Allowed grades are only Strong evidence, Moderate evidence, Plausible but uncertain, and Mostly coaching practice. A strong mechanism does not validate an exact prescription. Acute performance is not the same as a demonstrated training adaptation. Effect sizes, sample sizes, and protocol details below are taken from retrieved abstracts or, where noted, open full texts. If a number was not in the retrieved source, it is not stated. Many important trials remain abstract-only.

Evidence was searched through 2026-10-02. Papers with a first-publication date after that day were excluded. Europe PMC was the primary retrieval route. Web search was unavailable at the start of this investigation and was used after access was restored for original-paper discovery and 2025–2026 update scans. Search summaries were not used as quantitative evidence unless the abstract or full text was then retrieved.

Intensities do not adapt exclusive energy systems. Carbohydrate availability changes whether planned work can be completed and repeated. It does not create a separate adaptation system beside the training.

Durable conclusions

FUEL-C01 — Prolonged rides benefit from carbohydrate; 100–120 g/h is optional, not mandatory

Classification: Strong evidence that carbohydrate beats placebo as exercise gets longer. That Strong grade does not cover high-dose superiority. Moderate evidence for a broad hourly target (about 30–60 g/h, up to about 90 g/h when the ride is long). Plausible but uncertain that ≥100 g/h beats 60–90 g/h on a single cycling effort.

Support: Stellingwerff and Cox (2014) reviewed 61 randomised carbohydrate-versus-placebo performance studies (n = 679): 82% found a benefit, and the percent gain correlated with duration. Burke et al. (2011): small amounts or a mouth rinse near 1 hour; 30–60 g/h for longer work; up to 90 g/h beyond about 2.5 hours with mixed carbohydrates. Jeukendrup (2014, open full text) stated the same ceilings and said advice should be lowered when absolute intensity, and therefore carbohydrate oxidation, is low. Smith et al. (2013): 51 cyclists and triathletes, 2 hours plus a 20 km time trial, 0–120 g/h of a glucose–fructose–maltodextrin drink. Curvilinear response; best mean performance near 78 g/h, with diminishing returns above that. Wilson (2026, open full text) reviewed ≥100 g/h fueling in professional cycling and concluded that the small set of experiments versus 60–90 g/h does not clearly show less endogenous carbohydrate use or better single-session performance.

Contradiction / limits: Norte et al. (2026, open full-text results): 16 endurance-trained cyclists/triathletes (VO₂max 51.8 ± 6.8 mL·kg⁻¹·min⁻¹), carbohydrate-loaded, 3 hours at 95% of gas-exchange threshold, then a 3-minute all-out test. Fresh end-test power 277 ± 27 W; after water 236 ± 30 W; after 60 g/h 257 ± 28 W; after 120 g/h of 1:0.8 maltodextrin:fructose 266 ± 29 W. W′ did not differ between fatigued conditions. That is durability versus water and 60 g/h, not versus 90 g/h, and the preload was moderate laboratory cycling. King et al. (2018) did not find a linear time-trial benefit from 60 to 112.5 g/h; 90 g/h of glucose–fructose was the most favourable condition in that crossover of ten trained men.

Practical implication: Match intake to duration and to how hard the absolute work is. Most 7–12 hour weeks do not require 90–120 g/h. Use that range only for long hard rides or races, and only after it has been practiced. High-dose fueling is opt-in, not a new requirement.

Uncertainty: Dose trials are mostly in men. Laboratory time trials are not Grand Tours. Wilson’s case for ≥100 g/h across multi-day racing is theoretical and observational, not an amateur RCT.

FUEL-C02 — Mixed glucose or maltodextrin plus fructose beats glucose alone at high intake; the exact ratio is a smaller decision

Classification: Strong evidence that adding fructose raises exogenous oxidation above the single-source ceiling when intake is high. Moderate evidence that a ratio near 1:0.8 glucose:fructose is better than 2:1 for oxidation efficiency or comfort. Mostly coaching practice if the claim is that the ratio must be exact.

Support: Jeukendrup (2014, open full text) names SGLT1 saturation near 60 g/h of glucose. Podlogar and Wallis (2022, open full text) name GLUT5 as the fructose transporter and state that mixtures are recommended beyond about 2.5 hours at 60–90 g/h. They also argue that a ratio near 1:0.8 is superior for oxidation efficiency and gut comfort. Wallis et al. (2005): eight trained cyclists, 150 minutes, maltodextrin plus fructose at 1.8 g/min, peak exogenous oxidation 1.50 ± 0.07 g/min versus 1.06 ± 0.08 g/min for maltodextrin alone. Currell and Jeukendrup (2008): eight trained men (VO₂max 64.7 ± 3.9 mL·kg⁻¹·min⁻¹), 2:1 glucose:fructose at 1.8 g/min, time trial 8% faster than isocaloric glucose and 19% faster than water. The abstract’s completion times conflict with that direction, so the times are not used. O’Brien and Rowlands (2011): ten cyclists, 1.8 g/min, fructose:maltodextrin ratios 0.5, 0.8, and 1.25; the 0.8 ratio had a small oxidation advantage and the least fullness, cramping, and nausea. O’Brien et al. (2013): twelve male cyclists; the 0.8 ratio had higher exogenous oxidation efficiency and moderately higher sprint power than 0.5 and 1.25; fluid-absorption effects were inconclusive. Rowlands et al. (2015) reviewed 14 studies of about 2.5–3 hour performance, mostly male cycling, and estimated small-to-moderate mean-power gains (1–9%) for mixed drinks versus isocaloric glucose or maltodextrin at 1.3–2.4 g/min.

Contradiction / limits: Podlogar et al. (2022): eleven highly trained men (VO₂peak 62.6 ± 7 mL·kg⁻¹·min⁻¹), 3 hours at 95% gas-exchange threshold. Exogenous oxidation in the last hour was 1.51 ± 0.22 g/min at 120 g/h (0.8:1 fructose:maltodextrin) versus 1.29 ± 0.16 g/min at 90 g/h (1:2). Endogenous carbohydrate oxidation did not differ (2.15 ± 0.30 versus 2.20 ± 0.33 g/min). Higher oxidation of the drink did not spare body carbohydrate. Jeukendrup and Moseley (2010) showed faster gastric emptying and fluid delivery with glucose plus fructose than with glucose alone at 1.5 g/min in eight men. That is delivery, not proof that oxidation equals absorption. Ratio trials are small, mostly male, and several used magnitude-based inference.

Practical implication: For rides long enough to need more than about 60 g/h, use a glucose or maltodextrin source plus fructose. A 2:1 mix and a 1:0.8 mix are both defensible. Sucrose can be one ingredient in that mix; it has not been shown, taken alone, to match a free glucose-plus-fructose drink. Do not treat a 0.05 shift in ratio as a performance intervention. Do not assume a higher oxidation rate means less muscle-glycogen use.

Uncertainty: Females, heat, and race intensity are under-represented. Rowlands (2015) flagged that gap. Podlogar and Wallis (2022) suggest a near-unity ratio even for shorter sessions. That is a hypothesis from oxidation efficiency, not strong evidence that a 90-minute amateur ride requires fructose.

FUEL-C03 — Glucose plus sucrose, not sucrose alone, matched a free 2:1 mix for oxidation; table sugar is still not a sports-drink clone

Classification: Moderate evidence that a glucose-plus-sucrose drink supplying the same overall 2:1 glucose:fructose as free glucose plus fructose oxidises similarly in trained men. Moderate evidence, separately, that sucrose versus glucose (not versus a free mix) protects liver glycogen during moderate cycling and speeds liver, not muscle, glycogen repletion. Insufficient evidence, bordering on Mostly coaching practice, that sucrose alone matches a free glucose-plus-fructose drink, or that household sugar water matches a formulated product for gut comfort, sodium, fluid, or teeth.

Support: Trommelen et al. (2017; PMC5331598): ten trained male cyclists (VO₂peak 65 ± 2 mL·kg⁻¹·min⁻¹), 180 minutes at 50% Wmax, 1.8 g/min (108 g/h). Peak exogenous oxidation was 0.96 ± 0.06 g/min for glucose alone (1.8 g/min), 1.40 ± 0.06 for 1.2 g/min glucose plus 0.6 g/min fructose, and 1.29 ± 0.07 for 0.6 g/min glucose plus 1.2 g/min sucrose. That sucrose arm is not sucrose alone. After hydrolysis, 1.2 g/min sucrose is 0.6 g/min glucose plus 0.6 g/min fructose, so with the added 0.6 g/min glucose the drink is 1.2 g/min glucose plus 0.6 g/min fructose overall — the same 2:1 ratio as the free mix. Those two arms were similar and about 46% above glucose alone. Sucrose alone was not in that comparison. Jentjens, Venables, and Jeukendrup (2004): nine trained men; glucose plus sucrose (1.2 + 0.6 g/min), not sucrose alone, peaked at 1.25 ± 0.07 g/min versus about 1.06 g/min for glucose or glucose plus maltose. Jentjens, Achten, and Jeukendrup (2004): glucose plus fructose plus sucrose at 2.4 g/min reached about 1.70 g/min. Separately, Gonzalez et al. (2015): fourteen cyclists, 3 hours at 50% peak power, glucose or sucrose at 1.7 g/min (102 g/h). Liver glycogen did not fall with either sugar; muscle glycogen fell similarly; whole-body carbohydrate use was higher with sucrose (2.03 ± 0.43 versus 1.66 ± 0.36 g/min). In four cyclists who drank only water, both liver and muscle glycogen fell. Fuchs et al. (2016): fifteen well-trained male cyclists, 1.5 g·kg⁻¹·h⁻¹ sucrose versus glucose for 5 hours after depleting exercise. Muscle glycogen rose similarly; liver glycogen rose faster with sucrose (about 3.4 g/h; 95% CI 1.6–5.1). Those liver comparisons are sucrose versus glucose, not sucrose versus a free glucose-plus-fructose mix.

Contradiction / limits: These are moderate-intensity laboratory rides, not races. Jentjens et al. (2005) found that a high glucose-plus-sucrose intake (2.4 g/min) peaked near 1.20 g/min in that protocol, not at the highest rates seen in other studies. High intake is not automatically high oxidation. No retrieved trial showed sucrose alone matching free glucose plus fructose for exogenous oxidation, and none compared a kitchen sucrose solution with a commercial gel on symptoms, sodium balance, or dental erosion. A very concentrated sugar solution raises osmolality. A sugar-only bottle has no sodium.

Practical implication: Table sugar is a cheap, legitimate carbohydrate: hydrolysis yields glucose and fructose, and sucrose versus glucose has the liver findings above. It is not a proven stand-in, taken alone, for a free glucose-plus-fructose drink, and it is not a proven identical product. If sugar water is used, concentration, sodium, and volume still have to be chosen, and the mouth still sees sugar. Start with a concentration the gut already tolerates. Do not sip it all afternoon.

Uncertainty: The Trommelen oxidation match is open-access and is glucose plus sucrose at an overall 2:1, not sucrose alone. Gonzalez and Fuchs remain abstract-only. Gastrointestinal symptom data are not consistently reported in the abstracts used here.

FUEL-C04 — Oxidation is not absorption, and absorption is not tolerance

Classification: Strong evidence that the three are separable. Moderate evidence that short gut-training blocks can reduce symptoms in some runners. Mostly coaching practice if the claim is that two weeks of practice reliably unlocks 120 g/h and better cycling performance.

Support: Hearris et al. (2022): nine trained men, 120 g/h of 1:0.8 maltodextrin or glucose to fructose as drink, gel, chew, or a mix, after a high-carbohydrate day. Peak exogenous oxidation was about 1.56–1.66 g/min and did not differ by form; symptoms were minimal; a subsequent capacity test did not differ. Costa et al. (2017): 25 endurance runners, 2 hours at 60% VO₂max with 30 g carbohydrate every 20 minutes, then two weeks of gut training. Symptoms fell about 60% with gel or food training, and a distance test improved about 4–5% in those arms and not with placebo. Martinez et al. (2023) systematically reviewed eight gut-training or feeding-challenge studies: discomfort fell in some two-week carbohydrate protocols, malabsorption fell in two studies and not in one, and several symptom results were unclear. Gastric emptying and intestinal-injury markers generally did not change.

Contradiction / limits: King et al. (2022): two weeks of a very-high-carbohydrate diet plus gut training, then 90 g/h, in elite male race walkers and highly trained runners. Individual variability, more upper-gut symptoms in one cohort, and no clear performance advantage over strategies inside existing guidelines. Confounding weight loss and fitness gains were acknowledged. Not a cycling RCT. Cox et al. (2010): 28 days of training with high daily carbohydrate availability increased exogenous glucose oxidation in trained cyclists or triathletes, without a larger performance gain than the energy-matched lower-carbohydrate group (both improved about 6%). Practice can change oxidation. It does not guarantee a bigger training effect. Podlogar et al. (2026): twelve trained men ingested the same 90 g/h in increasing, decreasing, or constant patterns across 180 minutes plus a 15-minute test. No clear performance difference. A decreasing pattern had higher carbohydrate oxidation than constant intake and less late nausea than an increasing pattern. Even intake remains a sound default.

Practical implication: Practice the rate the event actually needs, on rides that resemble it. If 100–120 g/h is being considered, practice it and stop if symptoms are severe. Do not debut a new rate in a race, and do not assume the extra 30 g/h buys muscle-glycogen sparing.

Uncertainty: Gut-training trials are few, heterogeneous, and mostly running. Symptom tools differ.

FUEL-C05 — Fueling protects session quality and next-day work more clearly than it builds a separate adaptation

Classification: Strong evidence that low carbohydrate availability can reduce what a trained athlete can do in the session. Moderate evidence that carbohydrate during long work improves next-day performance. Plausible but uncertain that extra carbohydrate, once the session is well fueled, adds mitochondrial or performance adaptation beyond the training itself.

Support: Hulston et al. (2010): fourteen well-trained cyclists, 3 weeks. The low-glycogen group did the high-intensity session 1 hour after a 90-minute ride and produced lower interval power (297 ± 8 W versus 323 ± 9 W). Fat oxidation rose; time-trial performance was not better than training with high glycogen. Yeo et al. (2008): fourteen endurance-trained cyclists/triathletes, 3 weeks, same design family. High-intensity capacity was compromised in the twice-every-second-day group; selected mitochondrial markers rose only in that group; a 1-hour time trial after 60 minutes of steady riding improved about 10% in both groups. Hearris et al. (2019): eight men, sleep-low model. Starting below about 300 mmol/kg dry weight, further glycogen reductions of about 100 mmol/kg cut exercise capacity at 80% peak power by about 20–50% and did not further augment the measured cell-signalling response. Main et al. (2026): nine well-trained female cyclists, randomised crossover. After glycogen-depleting exercise, next-day carbohydrate was ≥9 g/kg or ≤1 g/kg. The original abstract reports power at the first ventilatory threshold of 133 ± 24 W (low) versus 152 ± 28 W (high), difference −19 ± 14 W, lower gross efficiency in the low condition, and no difference in lactate-threshold power or 3-minute all-out end-test power. Those wattage figures are original-abstract extracts only. The correction (doi:10.1007/s00421-026-06340-4; Springer full text retrieved) says: “In the original version of this article, Figure 2 was included twice as Figure 1 and Figure 2.” The replacement Figure 1 caption is a study schematic (High versus Low carbohydrate crossover, trials 5–14 days apart). The correction text does not mention watts, ventilatory threshold, or efficiency, so it neither revises nor confirms the −19 ± 14 W abstract result. This is acute durability of the moderate-to-heavy transition, not a training study. Clauss et al. (2023): seven trained male cyclists (VO₂max 66.8 ± 1.9 mL·kg⁻¹·min⁻¹), matched exhaustive work, 1 g·kg⁻¹·h⁻¹ carbohydrate versus water during the ride, isoenergetic diet otherwise. Next-day time trial was 41:49 ± 1:38 versus 42:50 ± 1:46 (small effect). Achten et al. (2004): seven trained runners, intensified training, 8.5 versus 5.4 g·kg⁻¹·day⁻¹. The lower intake worsened 16 km outdoor performance and mood more. Running, not cycling, and n = 7, but it is a direct test of daily carbohydrate during a hard block.

Contradiction / limits: Cox (2010) changed exogenous oxidation with daily carbohydrate availability and did not change the size of the performance gain. Molecular upregulation is not a performance result. See FUEL-C06. Wilson (2026) also flags a tradeoff if high in-ride doses are applied to every training ride: possible suppression of fat oxidation and a smaller metabolic signal. That is a reason not to pour 120 g/h onto every easy ride, not a reason to fast the quality sessions.

Practical implication: Fuel the sessions that are supposed to be high quality, and fuel long rides if the next day also matters. Do not fast before VO₂ or threshold work in the hope of a better adaptation. An easy 60–90 minute ride in a fed amateur does not require a sports drink.

Uncertainty: The Main (2026) wattage contrast is one small female crossover and remains an original-abstract figure; the retrieved correction does not settle it. Clauss (2023) is seven men and about a 1-minute difference. Neither defines a universal pre-session gram target. Most oxidation trials are in men; Main (2026) is the retrieved female cycling exception and is acute, not longitudinal.

FUEL-C06 — Train-low can change signals and fat use; it does not reliably improve performance in trained athletes

Classification: Moderate evidence that periodised carbohydrate restriction has not reliably improved endurance performance in trained athletes. That is not Strong evidence of a causal zero or a universal absence of benefit. Gejl and Nybo (2021) reported SMD 0.17 (95% CI −0.15 to 0.49), P = 0.29: no demonstrated reliable average advantage, with a confidence interval that still permits a potentially meaningful positive effect. A benefit of train-low remains Plausible but uncertain. Moderate evidence for acute increases in selected transcripts and for higher fat oxidation. Strong evidence, in elite race walkers, that sustained ketogenic diets can worsen economy; transfer to amateur cycling is not one-to-one. Not fasting quality sessions is a risk-oriented practical choice, not proof of a null.

Support: Gejl and Nybo (2021) meta-analysed nine studies in athletes at or above VO₂max 55 mL·kg⁻¹·min⁻¹ (women) or 60 (men), with carbohydrate restriction at least three times per week for at least a week. Standardised mean difference 0.17 (95% CI −0.15 to 0.49), P = 0.29. Gejl et al. (2017): 26 male elite endurance athletes (VO₂max 65 mL·kg⁻¹·min⁻¹), 4 weeks, manipulation 3 days/week. VO₂max and 30-minute time-trial power rose about 5% in both groups; no superior effect of restriction. Prieto-Bellver et al. (2024): 17 well-trained cyclists (VO₂peak 70.8 ± 6.5 mL·kg⁻¹·min⁻¹), 5 weeks, 13 low-availability sessions versus high carbohydrate, energy-matched. Maximal lactate steady-state power rose in both; time to exhaustion, substrate oxidation, heart rate, and lactate did not favour periodisation. Diaz-Lara et al. (2025) meta-analysed 19 randomised studies of acute low versus high carbohydrate availability: PDK4, GLUT4, and UCP3 mRNA rose in early recovery; PGC-1α and TFAM did not overall. PGC-1α mRNA rose when the glycogen gap exceeded about 200 mmol/kg dry weight. That is signalling, not performance. Impey et al. (2018) framed a glycogen window that still allows the session to be completed, and noted that performance improved in only a minority of the studies they counted.

The untrained origin should stay labelled. Hansen et al. (2005): seven untrained men, 10 weeks of one-legged knee-extensor training, work matched. Time to exhaustion at 90% increased more in the low-glycogen leg; maximal workload did not differ. That design does not transfer cleanly to trained cyclists, as Yeo (2008) and Hulston (2010) already showed.

Contradiction / limits: Marquet et al. (2016) reported better submaximal cycling economy, supramaximal capacity, and 10 km running after 3 weeks of sleep-low in 21 triathletes (VO₂max 58.7 ± 5.7 mL·kg⁻¹·min⁻¹) at a matched 6 g·kg⁻¹·day⁻¹, and a 1-week cyclist study reported about +3.2%. Those positive trials sit inside the Gejl meta-analysis. The pooled interval crosses zero and also reaches effects that could matter, so it does not prove the true average is zero. Samples are small and diets are hard to blind. Louis et al. (2016), on the same 3-week triathlete model, found minimal effects on selected immune markers, upper-respiratory symptoms, and sleep. That argues against a large immune penalty in that specific protocol. It does not prove that chronic under-fueling is safe.

Chronic ketogenic eating is not one sleep-low day. Burke et al. (2017): elite race walkers, 3 weeks of intensified training, isoenergetic high carbohydrate (8.6 g/kg), periodised carbohydrate, or ketogenic high-fat. VO₂peak rose in all groups. High and periodised groups improved 10 km race-walk time (6.6% and 5.3%); the ketogenic group did not (−1.6%) and economy worsened at race-relevant speeds. Burke et al. (2021): 5–6 days of ketogenic eating in world-class male race walkers raised the oxygen cost of race speeds by about 5–8%; 24 hours of carbohydrate restoration did not restore performance. Prins et al. (2025) reported similar time-to-exhaustion after 6 weeks of high- versus very-low-carbohydrate diets in trained triathletes, and a 22% time-to-exhaustion gain from only 10 g/h carbohydrate on both diets, attributed to avoiding hypoglycaemia. Outcomes and populations differ from Burke, and the Prins abstract is an incomplete methods record. A 2026 paired commentary restates the disagreement rather than resolving it (Burke and Noakes; Noakes and Burke). Kripp, Feichter, and König (2025): 24 recreational male runners, 8 weeks, including 4 weeks at ≤50 g carbohydrate/day. Fat oxidation changed reversibly; overall performance did not differ between periodised, chronic low-carbohydrate, and high-carbohydrate arms. Recreational running is not trained cycling.

Practical implication: For a 7–12 hour cyclist, the scarce resource is quality, not an extra molecular signal. Keeping carbohydrate available for VO₂, threshold, and long hard rides is a risk-oriented choice: low availability can cut the power of the session, and a performance benefit of withholding carbohydrate has not been reliably shown. It is not proof that train-low never helps. An occasional easy ride with low intake is not harmful by default and is not a proven upgrade. Do not adopt a ketogenic diet to raise FTP or VO₂max.

Uncertainty: Positive sleep-low studies exist. The pooled estimate does not demonstrate a reliable average advantage, and it does not rule one out. A universal “no benefit” claim is uncertain. Sex-specific train-low performance trials in cyclists were not retrieved.

FUEL-C07 — Tight turnarounds need early carbohydrate; muscle glycogen may still be unfinished at 12 hours

Classification: Moderate evidence, from controlled recovery studies and open reviews, mostly in men. About 1.2 g/kg/h early is the intake at which muscle-glycogen resynthesis is described as maximised when rapid restoration is the goal. It is not a minimum after every ride and not an open-ended floor.

Support: Alghannam, Gonzalez, and Betts (2018): when rapid restoration is required, about 1.2 g·kg⁻¹·h⁻¹ is the rate at which muscle glycogen repletion and repeated endurance capacity are described as maximised; protein co-ingestion is the fallback when carbohydrate is at or below about 0.8 g·kg⁻¹·h⁻¹. The reviews phrase this as ≥1.2 g·kg⁻¹·h⁻¹. That symbol means the resynthesis ceiling in those studies, not a quota for every next day. Gonzalez et al. (2017): the same muscle ceiling; glucose–fructose or sucrose speeds liver repletion and reduces distress near an intestinal glucose-absorption ceiling of about 1.2 g/min. Fuchs et al. (2025): twelve well-trained male cyclists (VO₂peak 67 ± 5 mL·min⁻¹·kg⁻¹; Wmax 5.8 ± 0.7 W/kg). Exercise cut muscle glycogen about 64% and liver glycogen about 34%. With 10 g/kg carbohydrate (1.2 g/kg/h as sucrose for the first 6 hours, then meals), liver glycogen exceeded pre-exercise values by 6 hours; muscle glycogen was 55% of pre-exercise at 6 hours and 69% at 12 hours. Fasting left both stores depleted. The ISSN nutrient-timing text (Kerksick et al., 2017) states that glycogen stores are maximised by 8–12 g carbohydrate/kg/day. Parts of that open abstract are corrupted in the record used here, so only the intact 8–12 g/kg/day sentence is used.

Contradiction / limits: Sollie et al. (2023) is not used as quantitative proof. The retrieved abstract confirms the design — 13 highly trained male cyclists, a 90-minute morning session including a 60-minute time trial, then 5 hours of recovery comparing 1.2 g/kg/h carbohydrate with 0.8 g/kg/h carbohydrate plus 0.4 g/kg/h whey — but the results paragraph was not fully extracted. The paper title states similar performance; that title is not treated as an outcome. If the first session did not empty the tank, the second is not automatically a glycogen emergency. Naderi et al. (2025) agree that early carbohydrate amount and type dominate short-turnaround glycogen. Adjuncts such as caffeine and creatine in that review are outside this brief.

Practical implication: When the next hard session is soon and glycogen restoration is the goal, start carbohydrate early at about 1.2 g/kg/h for the first several hours, and include some fructose or sucrose if the amount is large. Do not treat that rate as a minimum after every ride. Do not expect muscle glycogen to be full the next morning after severe depletion. For a single daily session in a 7–12 hour week, the daily total and the next pre-ride meal matter more than a 30-minute window.

Uncertainty: About 1.2 g/kg/h is a resynthesis ceiling when rapid restoration is required, not a daily minimum. A 1.0–1.2 g/kg/h band is common coaching language; the retrieved reviews pin the maximising rate at about 1.2 g/kg/h, so the lower edge of that band is not separately verified here. Body-mass scaling assumes the athlete can eat that much.

FUEL-C08 — Endurance protein needs sit above the sedentary RDA; timing is secondary to the daily total

Classification: Moderate evidence. Indicator-amino-acid studies are small. Position stands agree on a range, not a single optimum.

Support: Jäger et al. (2017), ISSN protein stand: 1.4–2.0 g/kg/day for most exercising individuals; about 0.25 g/kg or 20–40 g per serving every 3–4 hours; endurance athletes should still prioritise carbohydrate. Kato et al. (2016): six male endurance-trained adults (VO₂peak 60.3 ± 6.7 mL·kg⁻¹·min⁻¹) after a 20 km run; indicator-amino-acid oxidation implied a recommended intake of 1.83 g/kg/day. Wooding et al. (2017): six active women, variable-intensity shuttle running, recommended intake 1.71 g/kg/day. That is not cycling. Witard, Hearris, and Morgan (2025): about 1.8 g/kg/day, possibly above 2.0 g/kg/day when carbohydrate is restricted or on some rest days; no compelling performance or glycogen benefit from adding protein during exercise when carbohydrate recommendations are met. Morton et al. (2026), UCI Sports Nutrition Project: daily protein at least 1.6–2.1 g/kg/day, and carbohydrate periodised to the session rather than chronically restricted. Gillen et al. (2019): eight endurance-trained men; phenylalanine oxidation was higher after low-carbohydrate availability, and the authors concluded that protein requirements rise. The abstract does not state a new gram target.

Contradiction / limits: Indicator-amino-acid estimates are short, small, and not performance trials. ISSN ranges were built largely around resistance and mixed exercise. Female cyclists are almost absent. A higher protein estimate is not a reason to displace carbohydrate on hard days.

Practical implication: About 1.6–2.0 g/kg/day from food, spread across meals, is a credible target for this audience. Use the top of that range, or slightly above, in a calorie deficit or during deliberate carbohydrate restriction. Do not put protein in the bottle as a substitute for carbohydrate on long rides.

Uncertainty: Per-meal endurance dose-response data remain thin, as Witard and colleagues note.

FUEL-C09 — Replace fluid to the sweat actually lost; do not out-drink thirst into hyponatremia

Classification: Strong evidence that exercise-associated hyponatremia is usually caused by drinking beyond sweat and urine losses, with non-osmotic antidiuretic hormone release. Moderate evidence for individualised fluid and sodium targets. No universal sodium dose is justified.

Support: Sawka et al. (2007), ACSM fluid position: start euhydrated; limit excessive dehydration, described as more than about 2% body-weight loss from water deficit; customise replacement because sweat rate and sweat electrolytes vary; estimate sweat rate from body mass. McDermott et al. (2017), NATA: both under-replacement and overdrinking cause harm; quantify sweat rate; educate against hyponatremia. Hew-Butler et al. (2017): overdrinking beyond thirst and non-osmotic arginine vasopressin release are the most common causes; sodium depletion may matter in a subset; the syndrome is not confined to ultramarathons. Baker (2017) reviewed why sweat-rate and sweat-sodium tests mislead if the method is sloppy, and why intensity, heat, acclimation, body size, sex, diet, and hydration status all move the numbers. Barnes et al. (2019): observational tests in 1,303 athletes. Endurance sports averaged 1.28 ± 0.57 L/h whole-body sweat rate and 51.7 ± 27.8 mmol/h sodium loss. The standard deviations are the finding. This is not a cycling-only sample and not a prescription. Jeukendrup and Moseley (2010), cited under FUEL-C02, is the fluid-delivery contrast between glucose and glucose plus fructose.

Contradiction / limits: The 2007 “avoid >2% loss” goal and later “do not drink past thirst” advice can be pulled into a false conflict. They agree on individualisation and disagree with both “drink nothing” and “drink as much as possible.” A programmed 2% limit is not a reason to force fluid on a cool 90-minute ride. Barnes’ means must not be copied as a sodium tablet dose. The 2015 hyponatremia consensus statement was identified but its abstract text was not retrieved, so it is not used for numbers.

Practical implication: Weigh before and after a long or hot ride, corrected for drink, food, and urine. Replace most of the deficit during and after. Include sodium in long-ride fluid or food if sweat is salty and the ride is long, because losses are real and variable, not because a forum dose is correct. Never drink so much that body mass rises during exercise. This brief does not treat cramp and does not describe hyponatremia treatment. Confusion, vomiting, or bloating with weight gain during a long event needs urgent care.

Uncertainty: Sex-specific cycling sweat samples were not retrieved as primary data. A drink sodium concentration in mmol/L was not re-derived here.

FUEL-C10 — Low energy availability can limit health and training; a single calorie cutoff is not a diagnosis

Classification: Strong evidence that problematic low energy availability harms health in female and male athletes, and that diagnosis is clinical rather than a single energy number. Moderate evidence that under-fueling can erase or mimic a programming failure in specific overload studies. Insufficient evidence to assign a percentage of “recovery limited by intake versus programming.”

Support: Mountjoy et al. (2023) define REDs as deleterious health and performance outcomes from low energy availability in female and male athletes, distinguish adaptable from problematic exposure, and introduce Clinical Assessment Tool version 2 for detection and diagnosis by accumulated severity, not by one laboratory value. Stellingwerff et al. (2023) state that there is no singular validated diagnostic test. CAT2 is a three-step process: initial screening, severity stratification from primary and secondary indicators, then a physician-led diagnosis and plan. The traffic-light participation guide is clinical risk stratification, not a calorie cutoff. Areta, Taylor, and Koehler (2021) note that most causal low-energy-availability experiments last 5 days or less, that the concept was built first in women, and that male experimental evidence is limited. Lieberman et al. (2018): 35 sedentary ovulatory women, three menstrual cycles. The odds of a menstrual disturbance fell as energy availability rose (odds ratio 0.91 per unit, 95% CI 0.84–0.98), and no threshold emerged. Tertiles spanned roughly 23–50 kcal·kg FFM⁻¹·day⁻¹. This population is not trained cyclists; it is direct evidence against a single cutoff even in the group where the idea was developed. Logue et al. (2020) and Melin et al. (2019) review imperfect screening, including LEAF-Q in female athletes, and the fact that athletes can be weight-stable and still functionally impaired. Reported prevalence in track and field of about 18–58%, highest in endurance and jump events, is not cycling incidence. Stellingwerff et al. (2021): in 18 of 21 training-overload studies that also measured intake, there were signs of reduced energy availability or carbohydrate availability. Overtraining symptoms and REDs symptoms overlap. That does not mean every fatigued cyclist has REDs. It means under-fueling must be excluded before a programming diagnosis is trusted.

The Mountjoy (2023) full text was not retrieved (publisher access blocked). A widely circulated figure near 30 kcal·kg FFM⁻¹·day⁻¹ is therefore not quoted from that paper here. The retrieved CAT2 statement and the Lieberman trial are enough to reject using any single energy number as a diagnosis.

Contradiction / limits: Short laboratory energy-availability studies do not map onto free-living measurement error. Screening tools have false positives and false negatives. Weight loss is sometimes an athletic goal; the harm is the physiological cost, not the fact of a lighter rider. This brief does not set a minimum calorie intake or a minimum body-fat percentage.

Practical implication: If performance stalls, illnesses cluster, libido or menses change, bones hurt, or mood and recovery collapse, assess intake against the actual training load before adding another interval type. Screening is a prompt to involve a physician and, where available, a sports dietitian. It is not a self-diagnosis and not a reason for a universal calorie floor. For 7–12 hour riders, the common failure mode is hard sessions plus an aggressive weight-loss diet, not failure to copy a 30-hour professional’s bottle.

Uncertainty: Male cyclist REDs trials with performance endpoints were not retrieved as primary studies. The fraction of stalled athletes whose limiter is energy rather than program design is unknown. Do not invent that percentage.

13. Fueling and carbohydrate availability

Dose, duration, and intensity

Carbohydrate need scales with how long the work lasts and with how much carbohydrate the working muscle is actually oxidising, not with a badge on the bottle. Burke (2011) and Jeukendrup (2014) are the clearest retrieved statements of that scaling. Stellingwerff and Cox (2014) supply the performance evidence against placebo: benefit is common, and longer exercise shows larger percent gains. Near 1 hour, the mechanism they emphasise is oral exposure, not oxidation. Beyond about 2 hours, delivery and oxidation matter.

Situation Working range from retrieved guidance and trials What not to do
Hard session under about 75 minutes, athlete already fed Small amounts or a mouth rinse can help; the oxidation ceiling is irrelevant Do not force 90 g/h
About 1–2.5 hours, moderate to hard About 30–60 g/h is the Burke/Jeukendrup band; a single glucose source can cover the lower part Do not copy an ultra recipe
Beyond about 2.5 hours, or a long race Up to about 90 g/h of mixed carbohydrate if tolerated Do not assume more is linearly better
100–120 g/h Opt-in and practiced. Norte (2026) supports better critical-power preservation than 60 g/h or water after 3 moderate hours. Wilson (2026) finds no clear single-session win versus 60–90 g/h Do not make it the amateur default
Easy endurance, especially if the day is short and the athlete ate Water, or a small amount if the ride is long enough to drift Do not under-eat the day and then call the ride train-low by accident

Smith (2013) is the cleanest single dose-response in cyclists: curvilinear, best mean near 78 g/h across 0–120 g/h, with 1 L/h fluid. The sample was 51 cyclists and triathletes, mean age 28, not characterised as elite. King (2018) likewise failed to show a linear gain from 60 to 112.5 g/h at a hard 77% VO₂max preload. Wilson (2026), quoting the 2016 ACSM/AND/DC position stand, reports that events of 2.5 hours or longer are associated with better performance at intakes up to 90 g/h. That sentence is used as quoted by Wilson; the position-stand PDF was not opened independently.

Ijaz et al. (2025) complicate the older statement that during-exercise advice is independent of body weight. Fifteen cyclists ingested 90 g/h of glucose. Smaller athletes oxidised 33 ± 8 g/h exogenous glucose; larger athletes at the same relative intensity oxidised 45 ± 13 g/h (mean difference 13 g/h, 95% CI 2–24). Body surface area correlated strongly with peak exogenous glucose oxidation (r = 0.85). Group sizes in the abstract are incompletely specified. The direction is enough: a lighter rider and a heavier rider are not the same oxidation problem, and absolute intensity may be part of the reason. Jeukendrup (2014) said advice is independent of body weight. That sentence should now be treated as outdated for glucose oxidation rate, not as a refutation of duration-based ranges.

Transporters, ratios, and why oxidation is not absorption

SGLT1 and GLUT5 are named in retrieved full text (Jeukendrup 2014; Podlogar and Wallis 2022), not inferred. Glucose and maltodextrin share a saturable route. Fructose uses another. Maltose does not, which is why Jentjens, Venables, and Jeukendrup (2004) found glucose plus maltose no better than glucose, and glucose plus sucrose better. In that trial sucrose was added to glucose, not ingested alone. It helps because hydrolysis supplies fructose, not because it is a branded molecule. That is not evidence that sucrose alone matches a free glucose-plus-fructose drink.

Absorption, oxidation, and tolerance still diverge. A drink can leave the stomach, fail to be absorbed, be absorbed and not oxidised, or be oxidised and still cause distress. Podlogar (2022) increased exogenous oxidation from 90 to 120 g/h and did not reduce endogenous carbohydrate oxidation at a low absolute intensity after an overnight fast. Hearris (2022) showed that 120 g/h can be oxidised at similar rates from drink, gel, or chew in nine trained men who had carbohydrate-loaded, with little distress. Those are not the same athletes or the same feeding history. A number on a gel packet is an intake rate. The oxidation rate is lower, and the difference can become gut residue.

O’Brien’s 0.8 fructose:maltodextrin ratio is the best-retrieved primary evidence that 1:0.8 glucose:fructose can edge 2:1 (which is a 0.5 fructose:glucose ratio) for comfort and, in one study, sprint power. Effects were small to moderate, and the samples were 10–12 cyclists. Podlogar and Wallis (2022) go further and suggest using a near-unity ratio even for shorter sessions. That is a reasonable hypothesis from oxidation efficiency. It is not strong evidence that a 90-minute amateur ride requires fructose. Rowlands (2015) limited the performance inference to about 2.5–3 hour efforts in men.

Sucrose and ordinary table sugar

The metabolic trials are real and are in trained cyclists. They are summarised in FUEL-C03. Read them for what they measured.

  • Exogenous oxidation: Trommelen (2017) compared 0.6 g/min glucose plus 1.2 g/min sucrose (overall 2:1 glucose:fructose after hydrolysis; 108 g/h) with 1.2 g/min glucose plus 0.6 g/min fructose, in ten trained men. Oxidation was similar, and both beat glucose alone. Sucrose alone was not tested in that comparison. Jentjens (2004) likewise used glucose plus sucrose, not sucrose alone.
  • During exercise, liver glycogen: both glucose and sucrose at 102 g/h prevented net depletion; muscle glycogen fell either way (Gonzalez 2015). Carbohydrate feeding at this dose is liver protection and extra fuel, not muscle-glycogen sparing.
  • After exercise, liver glycogen: sucrose faster than glucose; muscle similar if the dose is high (Fuchs 2016; Gonzalez 2017).

What they did not measure is sucrose alone versus a free glucose-plus-fructose drink, or the amateur product question: a bottle of table sugar and water, no sodium, homemade concentration, versus a commercial drink or gel, on a hard outdoor ride, for gut symptoms, dental acid, and whether the rider finishes the bottle. Ordinary sugar can be the carbohydrate. It cannot be assumed to be the whole product, and it cannot be assumed, taken alone, to match the oxidation of a formulated glucose-fructose mix.

Gut tolerance and gut training

Distress is a reason to change rate, concentration, or form, not a character flaw and not proof that the physiology needs 120 g/h. Martinez (2023) is the right grade for gut training: possible reductions in discomfort and malabsorption, inconsistent symptom results, little change in gastric emptying or injury markers, eight small studies. Costa (2017) is the most cited positive running trial and should not be generalised to “every cyclist can learn 120 g/h in a fortnight.” King (2022) is the caution: highly trained athletes, short very-high intake, minor or absent performance gain, and sometimes more symptoms.

Fluid, sodium, and hyponatremia

See FUEL-C09. For a 7–12 hour rider the usual risk on a cool short ride is not hyponatremia. The usual risk on a hot long ride is either drinking nothing or copying a schedule that exceeds sweat rate. Measure one or two representative rides. Include sodium in long-ride fluid or food because sweat sodium losses are real and variable (Barnes 2019; Baker 2017), not because a forum dose is correct. If body mass is rising, stop drinking ahead of thirst.

Dental health

Elite athletes often have poor oral health. Needleman et al. (2015) list nutrition, mouth dryness, immune change, and low prioritisation of dental care as challenges, and note that the research base is limited. Ashley et al. (2015) systematically reviewed elite athletes: caries prevalence about 15–75%, erosion about 36–85%, periodontal disease about 15%, and 5–18% of athletes in four studies reported an impact on performance. Methodological quality was generally low. The London 2012 dental-clinic sample (Needleman 2013) found caries in 55%, erosion in 45%, and 18% reporting an impact on training or performance, among athletes who attended a clinic. That is a selected sample. Schulze and Busse (2024) attribute part of the burden to frequent sugar and carbohydrate exposure and argue for hygiene alongside the sports diet, not instead of fueling.

Practical reading: fuel the ride that needs carbohydrate. Do not sip sugary or acidic drinks across the rest of the day. Rinse with water after bottles and gels. Keep ordinary dental care. Do not drop in-ride carbohydrate on long rides to protect enamel; the performance and recovery cost is the larger, better-measured risk. Do not claim that sports drinks are the proven cause of the prevalence figures above. The studies do not isolate that exposure.

Training quality versus structural adaptation

The useful distinction is acute versus longitudinal.

Acute: low carbohydrate availability reduces interval power (Hulston 2010), can cut capacity once glycogen is already low (Hearris 2019), lowers power at the moderate-to-heavy transition and efficiency in trained women (Main 2026), and, when carbohydrate is omitted during exhaustive cycling, slightly worsens next-day time-trial performance (Clauss 2023). Norte (2026) shows the same direction for critical power after 3 hours: more carbohydrate, less drop, still a drop. These are session and next-day effects.

Longitudinal: doing some sessions low can increase fat oxidation and some transcripts (Yeo 2008; Diaz-Lara 2025) without a better race result (Gejl 2021; Prieto-Bellver 2024). Daily high carbohydrate during training can increase exogenous oxidation (Cox 2010) without a larger performance gain than an energy-matched lower intake in that 28-day study. The adaptation that matters for VO₂max and FTP still comes from the training that was completed and recovered. Carbohydrate is an enabler.

Train low, without the folklore

Strategy What the retrieved studies actually did Performance result in trained athletes
Twice a day, second session low Yeo 2008; Hulston 2010; trained cyclists, 3 weeks Markers or fat use up; interval power down; time trial not better
Sleep low, matched daily carbohydrate Marquet 2016; triathletes or cyclists, 1–3 weeks Positive in those small trials
Periodic restriction, elite Gejl 2017; 4 weeks No superior adaptation
Periodised versus high, well-trained cyclists Prieto-Bellver 2024; 5 weeks, VO₂peak about 71 No superior maximal lactate steady state or oxidation result
Meta-analysis of periodised restriction Gejl and Nybo 2021; 9 studies SMD 0.17 (95% CI −0.15 to 0.49). No demonstrated reliable average advantage; interval still allows a meaningful benefit. Not proof of zero
Ketogenic high-fat, elite race walk Burke 2017, 2021 Higher fat oxidation, worse economy, no race benefit
Untrained one-leg model Hansen 2005 Better time-to-exhaustion in the low leg; not a cycling prescription

Morton et al. (2026) restate the practical consensus: fuel for the work required, periodise carbohydrate to the session, and do not expect definitive performance superiority from train-low. Podlogar and Wallis (2022) reach the same performance conclusion from the other direction: high carbohydrate availability remains the priority for intense competition, and train-low has limited utility for performance despite its popularity.

14. Recovery nutrition and energy availability

Daily targets and timing

Target Range Source and limit
Carbohydrate to maximise glycogen stores 8–12 g/kg/day ISSN 2017 (Kerksick). Top of the range for heavy training, not an easy-day quota
Carbohydrate during intensified training 8.5 versus 5.4 g/kg/day Achten 2004. High intake better preserved running performance and mood. n = 7, running
High-carbohydrate arm in elite race walkers 8.6 g/kg/day Burke 2017. Not an amateur cycling RCT
During exercise 30–60 g/h, up to about 90 g/h beyond about 2.5 h Burke 2011; Jeukendrup 2014. Scale down if absolute intensity is low
Rapid muscle-glycogen restoration About 1.2 g/kg/h early, only when rapid restoration is the goal Alghannam 2018; Gonzalez 2017. Not a minimum after every ride
Protein, exercising adults 1.4–2.0 g/kg/day ISSN 2017 (Jäger)
Protein, endurance-oriented updates about 1.6–2.1 g/kg/day; point estimates near 1.8 Kato 2016 (n = 6 men); Witard 2025; Morton 2026
Protein per meal about 0.25–0.40 g/kg or 20–40 g, every 3–4 h ISSN 2017. Built more on muscle-protein synthesis than on cycling time trials

Timing matters when the next session is soon. It matters much less when the next hard session is a day away and the daily totals are met. That is the practical content of the nutrient-timing literature for this audience, not a 20-minute metabolic window.

Glycogen between closely spaced sessions

Fuchs (2025) is the clearest recent cycling result. Well-trained men, severe depletion, 10 g/kg in 12 hours: liver full within 6 hours, muscle still about a third down at 12 hours. Twice-daily professional training lives in that gap. A 7–12 hour amateur rarely does. When the amateur does ride hard in the evening and again the next morning, the evening carbohydrate intake is the intervention, not a new interval scheme. Sucrose or glucose plus fructose is the better liver tool (Fuchs 2016; Gonzalez 2017). Muscle still needs total carbohydrate and time.

Energy availability, sex, and health

Energy availability is a calculation with large error in free-living athletes (Areta 2021; Logue 2020). Low energy availability is a physiological exposure. REDs is the clinical syndrome. The 2023 tool separates screening from diagnosis on purpose (Stellingwerff 2023; Mountjoy 2023). See FUEL-C10.

Sex belongs in the interpretation, not as an afterthought.

  • Most exogenous-oxidation and train-low performance trials retrieved here are in men.
  • Main (2026) is the retrieved female cycling exception and is acute, not longitudinal. The original abstract reports a −19 ± 14 W drop at the first ventilatory threshold (133 ± 24 vs 152 ± 28 W) and lower efficiency with low carbohydrate availability, with no change in 3-minute all-out end-test power. The Springer correction (doi:10.1007/s00421-026-06340-4) fixes a duplicated figure and does not revise or confirm those watts. Fueling is not a male-only performance variable, but the wattage contrast remains abstract-only.
  • Menstrual disturbance is a clinical flag, not a vanity metric. Lieberman (2018) rejects a single energy-availability threshold even in sedentary women. Melin (2019) reports high rates of functional hypothalamic amenorrhoea in elite female endurance runners, with bone and cardiovascular concerns. Those prevalence figures are track and field, not a cycling incidence.
  • Male low energy availability can present as low testosterone, low libido, and poor training capacity (Melin 2019; Mountjoy 2023 abstract). Experimental male trials are fewer (Areta 2021).
  • Wooding (2017) is a reminder that female protein estimates exist and are not automatically the male number, and also that the study was shuttle running.

Health safety, without turning this into a medical manual: persistent fatigue, recurrent illness, stress fractures or bone pain, loss of menses, loss of libido, or disordered eating are reasons to stop self-experimenting with fasted training or aggressive deficits and to get clinical assessment. Iron is outside this brief. This page does not set a minimum body-fat percentage or a minimum calorie intake.

How much is nutrition, and how much is programming?

The literature does not support a fabricated percentage. It supports a qualitative rank.

Studies in which intake clearly changed the outcome of a training or overload block include Achten (2004; higher daily carbohydrate better preserved running performance during intensified training), Burke (2017; ketogenic adaptation removed the performance benefit of intensified training despite a higher VO₂peak), and Stellingwerff (2021; most overload studies that measured intake also found an energy or carbohydrate shortfall). Clauss (2023) and Norte (2026) show smaller, acute, next-session effects of in-ride carbohydrate. Gejl (2021) and Prieto-Bellver (2024) show that clever carbohydrate periodisation does not outperform simply fueling the same training.

For a 7–12 hour cyclist, the programming variables that still dominate are the recoverable dose of specific work: enough easy volume to support frequency, a small number of hard sessions done well, and progression. Nutrition becomes the limiter when those sessions are systematically under-fueled, when daily energy is too low for the week, or when two hard days are stacked without carbohydrate. It is not the limiter because the gel ratio was 2:1 instead of 1:0.8. Wilson (2026) is useful in the other direction: professional narratives that recent performances are mostly a carbohydrate revolution outrun the single-session trials. Both errors assign a precise share of performance to nutrition without a denominator.

7–12 hours versus 20–30 hours

Decision Trained amateur, 7–12 h/week 20–30 h/week or stage racing
Days that need 60–90 g/h The long ride and hard rides beyond about 2 hours. Often one or two days Several days; in-ride carbohydrate is a large fraction of daily carbohydrate
100–120 g/h Optional experiment for a long event, after practice. Not a weekly habit More plausible across multi-day racing for total intake and recovery (Wilson 2026), still not proven superior to 90 g/h on one stage
Train low Low value. Quality sessions are a large share of the week. Do not blunt them Still not a performance guarantee. Easier to hide one easy low-availability session inside a big week. Hard sessions and races stay fueled
Daily carbohydrate Hard days toward the high end of what the athlete can eat. Use 8–12 g/kg only when the load justifies it. Easy days lower, following fuel-for-the-work-required rather than an unretrieved easy-day gram table Consistently high. 8–12 g/kg is more often relevant. Under-eating shows up as failed sessions, illness, or REDs
Recovery timing Usually one session a day. Daily total dominates Twice daily and stage gaps make the first 4–6 hours material. Muscle glycogen may still be incomplete at 12 hours (Fuchs 2025)
Gut training Practice the rate you will race, once that rate is justified More relevant, still individual
Hyponatremia Mainly long, hot, or event days if drinking is excessive Longer exposure. A measured sweat plan matters more
Energy availability Risk rises if weight loss is stacked on hard training Risk rises because expenditure is enormous and appetite, travel, and body-composition pressure do not automatically match it

A 30-hour professional week is not a scaled-up 10-hour week. The amateur’s marginal hour is more often a quality problem or a recovery problem. The professional’s marginal hour is more often an energy-availability and serial-glycogen problem. Copying the bottle without copying the hours is how amateurs over-fuel easy rides and under-fuel the one session that needed carbohydrate.

Evidence capsules

Capsules are restricted to design, population, and results that were in a retrieved abstract or full-text passage.

Smith 2013. Four-site crossover, 51 cyclists/triathletes, 2 hours at 95% of the workload that elicited 4 mmol/L lactate, then 20 km time trial, 0–120 g/h. Curvilinear performance, estimated optimum 78 g/h. Abstract-only. Not elite-characterised.

Currell 2008. Eight trained men, 120 minutes at 55% Wmax then about 1 hour time trial, 1.8 g/min glucose versus 2:1 glucose:fructose versus water. Abstract conclusion: 8% faster than glucose, 19% faster than water. Abstract times conflict with that direction and are discarded. Abstract-only.

Gonzalez 2015 and Fuchs 2016. Trained cyclists, ¹³C magnetic resonance spectroscopy. During exercise, glucose and sucrose similarly protect liver and fail to spare muscle. After exercise, sucrose favours liver. Abstract-only. Men. Moderate intensity during the feeding trial.

Podlogar 2022 oxidation. Eleven highly trained men, 3 hours at 95% gas-exchange threshold, 120 versus 90 g/h, different ratios. Higher exogenous oxidation, no endogenous sparing, no performance test. Open access; abstract used for the numbers.

Norte 2026. Sixteen endurance-trained cyclists/triathletes, not elite by the reported VO₂max, carbohydrate-loaded, 3 hours at 95% gas-exchange threshold, 0 versus 60 versus 120 g/h, then 3-minute all-out. Dose-ordered critical-power preservation, no W′ difference. Full-text results paragraph. No 90 g/h arm.

Wilson 2026. Narrative review, professional cycling focus, open full text. Key point: existing research does not directly support ≥100 g/h versus 60–90 g/h for performance; studies do not mimic multi-day stage racing; risks include gut symptoms and, if over-applied in training, a smaller metabolic signal.

Gejl and Nybo 2021. Meta-analysis in trained endurance athletes. SMD 0.17 (95% CI −0.15 to 0.49), P = 0.29. No demonstrated reliable average performance advantage of periodised carbohydrate restriction. The interval permits a potentially meaningful positive effect, so this is not evidence that the true effect is zero. Abstract contains the pooled estimate.

Main 2026. Nine well-trained women, acute crossover, not a training study. Original abstract: ventilatory-threshold power 133 ± 24 W (low carbohydrate) versus 152 ± 28 W (high), difference −19 ± 14 W; efficiency lower in the low condition; lactate-threshold power and 3-minute all-out end power not different. Correction full text (doi:10.1007/s00421-026-06340-4): Figure 2 had been duplicated as Figure 1; replacement Figure 1 is the crossover schematic (5–14 days apart). No wattage change is stated. The −19 W figure is therefore original-abstract only, neither revised nor confirmed by the correction.

Fuchs 2025. Twelve well-trained male cyclists, 10 g/kg in 12 hours, liver restored, muscle not. Abstract-only.

Burke 2017 and 2021. Elite race walkers, not cyclists. Ketogenic adaptation raises fat oxidation and can impair economy and race outcome. Strong within that population. Do not silently transfer the percent changes to amateur cycling.

Open questions

  • Does 100–120 g/h beat 90 g/h on cycling performance when the comparison is fair on ratio, osmolality, and a race-like preload? Norte (2026) did not include 90 g/h. Wilson (2026) says the existing set does not show a clear win.
  • Do multi-day stage races, as opposed to one laboratory bout, reward higher in-ride intake through next-day glycogen? Plausible, untested as a randomised cycling trial in the sources retrieved here.
  • How should hourly targets change with body size and with sex? Ijaz (2025) is a start for glucose oxidation and body size. Female oxidation dose-response trials were not retrieved.
  • Does a 1:0.8 ratio beat 2:1 by a meaningful amount in women, in the heat, and in events under 2.5 hours? Podlogar and Wallis (2022) suggest yes for efficiency. Primary performance data do not settle it.
  • Is there a trained-cyclist trial of table sugar versus a commercial product on symptoms, sodium, and teeth? Not retrieved.
  • Will a newer meta-analysis overturn Gejl and Nybo (2021)? Diaz-Lara (2025) and Prieto-Bellver (2024) do not. They separate signalling from performance.
  • What energy-availability measurement, if any, is accurate enough for individual prescription? Not established. CAT2 is a clinical process, not a calculator.
  • How large is the share of stalled amateur progress that would reverse with energy rather than with a different interval? Unknown. Do not invent it.

Access note

Primary retrieval was Europe PMC (abstracts and selected PMC full texts) plus publisher or DOI records. Web search was unavailable at the start of this investigation and was used after access was restored for 2025–2026 discovery (Wilson 2026; Diaz-Lara 2025; Prieto-Bellver 2024; Norte 2026 confirmation; Podlogar 2026 feeding-pattern paper; Main 2026). Quantitative results come from retrieved abstracts or full text, not from search-engine summaries. Publications after 2026-10-02 were not sought. The Mountjoy (2023) full text could not be opened; the abstract and the companion CAT2 paper were used instead. The Main (2026) correction was retrieved from the Springer HTML full text: it corrects a duplicated figure and does not state a change to the ventilatory-threshold watts. Many classic trials (Currell, Smith, Gonzalez, Fuchs, Hulston, Yeo, Marquet, Burke race-walk papers) remain abstract-only in this record. Sollie (2023) results were not fully extracted; the title is not used as an outcome.

References

Links are DOI pages. Abstract-only unless the conclusion text says open full text.