REF-1475 · Jones et al., 2026 · Scandinavian Journal of Medicine & Science in Sports · Paper · PubMed
How much carbohydrate to load if I'm still riding lightly?
In 11 cyclists, 10 g/kg a day for 48 h stored more than 8 or 6; 8 did not beat 6. Gut: only fullness rose. Glycogen, not race time, after a hard session and a low-carb day.
Main caveat: Small sample: n = 11, with one missing biopsy at 8 g/kg/day; the power estimate came from a wider contrast and does not cover 6 vs 8, so "no difference" is not…
Jones and colleagues (Liverpool John Moores University, Nottingham Trent University) asked how much carbohydrate fills muscle glycogen when athletes keep training before a race, because most loading studies rest their participants.
Eleven endurance-trained cyclists (8 men, 3 women) completed three 5-day periods with all food supplied: about 36 h at 4 g/kg/day and 60 min of light riding, a hard interval session on day 2 that lowered glycogen by 71%, then 48 h at 6, 8 or 10 g/kg/day with 60 and 30 min of light riding, and a biopsy on day 5, 2 h after a 2 g/kg breakfast.
Glycogen after 10 g/kg/day (635.5 mmol/kg dry mass) was higher than after 8 (506.1) and 6 (460.9), which did not differ. Stomach fullness was the only gut symptom to rise; stool type, bowel frequency, gut transit time, body mass and body water did not differ. The authors conclude there is a linear dose-response with no saturation up to 10 g/kg/day.
Our reading: a direct dose comparison with training continuing, which most loading trials lack, but the claim it supports is narrower than the abstract's.
Ten beat eight on glycogen in 11 people after a run-in that lowered their stores; whether that extra glycogen wins races is untested, and this group's own audit (REF-1364) doubts how much of loading's performance benefit survives blinding. It is also one of the trials inside their meta-analysis (REF-1204), so the two are not independent support.
Protocol
Sample: n = 11 (8 men, 3 women; age 24 ± 5 years; 71.2 ± 12.0 kg; VO2max 56 ± 6 mL/kg/min; peak power 4.3 ± 0.4 W/kg; habitual carbohydrate 3.9 ± 1.0 g/kg/day). Of the women, one was naturally menstruating (per the title of Table S2; supplement not read).
Exercise, days 1-4: day 1, 60 min light-moderate steady cycling at about LT1 (mean 164 ± 49 W); day 2, 5 min warm-up, 30 min at about 65% VO2max, then 8 × 5 min at 80% of peak power with 1 min recovery; day 3, 60 min at the same light-moderate power; day 4, 30 min at that power. All at 1500 h in the laboratory; no other exercise allowed; none on day 5.
Diet: days 1-2 until the hard session, 4 g/kg/day carbohydrate; straight after it 1.2 g/kg carbohydrate and 20 g protein, then 2 g/kg carbohydrate 2 h later; days 3-4, 6, 8 or 10 g/kg/day with protein 2, fat 1 and fibre 0.5-0.6 g/kg/day; day 5 breakfast 2 g/kg carbohydrate. High and very high diets were topped up with zero-fibre products (jelly sweets, orange juice, sports drinks). No caffeine or alcohol; water ad libitum.
Compliance: all food pre-packaged and weighed; a photo of each meal sent by WhatsApp at the time of eating; containers weighed before and after for leftovers (97.1% ± 1.0% compliance). Actual intake is in the table below.
Glycogen: vastus lateralis biopsies by Weil-Blakesley conchotome under local anaesthetic, alternate legs, one on day 5 of each condition, 2 h after breakfast; two more on day 2 of the first condition only (baseline and after the hard session). Freeze-dried, acid hydrolysis, glucose kit; duplicate CV 10.5% ± 9.3%.
Gut and body: 0-10 symptom scales after exercise and before sleep on days 3-4 and before the biopsy on day 5; Bristol stool chart; blue-muffin transit time; body mass and eight-electrode bioimpedance water on the morning of day 5, fasted.
Analysis: linear mixed models (3 missing glycogen points, one at 8 g/kg/day post-load, estimated by maximum likelihood); Friedman and Bonferroni-adjusted Wilcoxon for symptoms; Pearson correlation as a descriptive supplement.
Funding and interests: funded by a Liverpool John Moores University Vice Chancellor PhD Scholarship to the first author; the authors declare no conflicts of interest.
Key Results
| Prescribed (g/kg/day) | Actual (g/kg/day) | Carbohydrate (g/day) | Energy (kcal/day) | Glycogen, mmol/kg dry mass (mean ± SD) | Gut transit (h) |
|---|---|---|---|---|---|
| 6 | 5.9 ± 0.1 | 419 ± 74 | 2863 ± 507 | 460.9 ± 100.7 | 20.2 ± 5.0 |
| 8 | 7.8 ± 0.2 | 560 ± 100 | 3427 ± 610 | 506.1 ± 124.0 | 19.0 ± 4.0 |
| 10 | 9.7 ± 0.3 | 691 ± 120 | 3940 ± 682 | 635.5 ± 78.0 | 18.7 ± 5.8 |
Pairwise glycogen: 10 higher than 6 and 8 (p < 0.03 in the abstract; p < 0.05 for all other conditions in the Results); 6 vs 8 p = 1.00; 4 vs 6 p = 0.20; 8 and 10 above the 4 g/kg/day baseline.
Change per 2 g/kg/day: 4 to 6, +107.1 ± 123.2; 6 to 8, +35.3 ± 118.8; 8 to 10, +117.8 ± 137.3 mmol/kg dry mass; the changes did not differ (p = 0.47, partial eta squared 0.09). The SD of each change exceeds its mean, with no repeat condition to separate individual response from biopsy noise. From these, with n = 10 pairs and unadjusted (our arithmetic), 95% CI about −50 to +120 for 6 to 8 and +20 to +216 for 8 to 10.
Correlation: r = 0.71 (relative) and r = 0.64 (absolute intake) with glycogen, both p < 0.001, across 4, 6, 8 and 10 g/kg/day and not accounting for repeated observations.
"Linear" against the pairwise results: the pairwise results neither show nor exclude a straight line. The steps were 107, 35 and 118, 6 and 8 did not differ, and only 10 stood above its neighbour; but a straight line from 6 to 10 would rise about 87 per step (our arithmetic), inside both confidence intervals above, so 6 = 8 is not evidence of a step.The claim rests on a correlation the authors call descriptive, which includes a baseline point measured once, after about 36 h rather than 48 h, and on a test that failed to find the steps different, in 11 people. No fit of a line against a step was reported. The data are a rising trend compatible with a line or a step at 10; they do not establish linearity.
Hard session: glycogen fell by 266.9 ± 79.4 mmol/kg dry mass (−71%).
Stomach fullness, median (IQR, 0-10 scale): 1 (0-1), 0 (0-2) and 2 (0-4) at 6, 8 and 10 (Friedman p = 0.006; 6 vs 10 p = 0.017, 8 vs 10 p = 0.014). Nausea, reflux, cramps, flatulence and urge to defecate did not differ (p from 0.097 to 0.779).
Moderate-severe scores: the text reports more scores of 5 or above at 10 g/kg/day for nausea, reflux, fullness and flatulence, without counts in the main text; one woman gave 55% of them.
Body: body mass on day 5 p = 0.70; total, extracellular and intracellular water p = 1.00, 0.25 and 0.58. The authors calculate that a 4:1 water-to-glycogen ratio should have added 0.12, 0.70 and 0.58 kg (6 vs 8, 6 vs 10, 8 vs 10), which was not detected, and call the measures possibly influenced by the procedures.
Other: hunger higher at 6 than 10 (p = 0.05), fullness on the scale higher at 10 than 6 (p = 0.02); fluid 2.1, 2.5 and 2.6 L across days 3-4 (p = 0.11); sodium 1751, 2223 and 3009 mg/day (retrospective, not controlled); carbohydrate in the 0-4 h after the sessions about 2, 3 and 4 g/kg.
In [REF-1204]: this trial is that meta-analysis's reference 51, with high and very high arms merged for the overall estimate and entered separately in the dose subgroups. Its Table 1 lists the sample as "11 M", while this paper reports 8 men and 3 women and the review's own text counts it among trials recruiting both sexes.
Numbers that do not add up: the Discussion's "172% increase" reads as 172% of baseline: −266.9 being −71% puts baseline near 376 mmol/kg, and 635.5 is about 1.7 times that, not 2.7 (our arithmetic).
Mechanisms
Measured: glycogen after 48 h rose with carbohydrate intake and was highest at 10 g/kg/day; carbohydrate in the 4 h after each light session was also highest at 10; energy and sodium rose with the dose, because fat and protein were held per kilogram.
Authors' interpretation: no plateau up to 10 g/kg/day, so the ceiling of 7-10 g/kg/day proposed from 24-h studies is yet to be determined; participants had about 60 h of raised carbohydrate intake (the re-feed plus the 48 h), against 24 h in Burke 1995, offered as why 10 g/kg/day stored more than that study's 11.8; continued training may explain a rise smaller than in resting studies; about 4 g/kg in the 0-4 h after exercise may have raised synthesis.
Authors' interpretation: the missing body-mass rise may reflect bioimpedance being affected by unequal electrolytes, small samples and a long protocol, and the 4:1 ratio itself is debated.
Inference: carbohydrate dose, energy, sodium and post-session timing all moved together, so this trial cannot say which of them produced the gain; there was no resting arm, so it does not measure what training costs. The arms ate 7.8 and 9.7 g/kg/day, both inside the 7-10 range the authors contrast with, so the trial shows no plateau at about 8 in this protocol, not that the range is wrong (our reading).
Practical Application
How much carbohydrate to load if I'm still riding lightly? In 11 cyclists, 10 g/kg a day for 48 h stored more than 8 or 6; 8 did not beat 6. Gut: only fullness rose. Glycogen, not race time, after a hard session and a low-carb day.
What the tested day looks like: for a 71-kg rider about 691 g of carbohydrate and 3,940 kcal, with fibre held near 0.5 g/kg by jelly sweets, juice and sports drinks. High-fibre loading was not tested.
Do not promise no weight gain: the scale and bioimpedance found no change, but the authors expected about 0.6-0.7 kg and doubt the measures could see it. This trial does not clear loading for a weight-sensitive event.
Not from this paper; the base's own practice: this does not change whether to load (events past about 90 min, REF-819, PHYS-04), the absence of a depletion phase, or the 1-3 days (SYNTH-64); from the paper: with easy sessions after a hard session and a low-carbohydrate run-in, about 8 stored less than about 10; a load from normal stores was not tested, nor any intake between 7.8 and 9.7.
What it does not show: a race or time-trial gain (REF-1364), anything above 10 g/kg a day, a load from normal stores without the hard session, gut comfort in the race itself, or elite response.
Limitations
Small sample: n = 11, with one missing biopsy at 8 g/kg/day; the power estimate came from a wider contrast and does not cover 6 vs 8, so "no difference" is not equivalence.
Outcome: thigh glycogen 2 h after a 2 g/kg breakfast; no performance test.
One biopsy per condition: duplicate CV 10.5% ± 9.3%; no repeat condition.
Run-in: about 36 h at 4 g/kg/day and a session cutting glycogen by 71% came first, so this is not a load from full stores and does not test a no-depletion regimen. Habitual intake (3.9 g/kg/day) was already low.
Confounded dose: energy, sodium and post-session carbohydrate rose with the dose.
Blinding: participants learned the diet on the first day of each condition; no blinding of participants or analysts is reported. Fullness, hunger and symptoms are self-reported.
Body measures: bioimpedance with unequal sodium intakes; the authors themselves doubt the sensitivity. The title's "body composition" means body mass and body water only.
Women: 3 of 11, one naturally menstruating; symptoms are reported as a total, and glycogen is not split by sex in the main text.
Population: trained amateurs at 7 ± 3 h a week; the authors say transfer to elites is undetermined.
Not read: supplementary Tables S1-S5 (contraception and cycle phase, energy availability, carbohydrate timing, individual body water).
Funding and interests: university PhD scholarship; no conflicts declared. A Science in Sport drink and bar were the post-session feed, with no support from the company stated.
How this entry was checked and changed
Why this level
crossover with n = 11 (rubric B, under 15), lowered a step because the outcome is a marker, muscle glycogen, not performance; no blinding reported, and the sample-size estimate (n = 4) was borrowed from a much wider intake contrast (2.4 to 6.5 g/kg)
Full text (checked 2026-10-30)
Days 1, 3 and 4, participants visited the laboratory at 1500 h and completed 60-, 60- and 30-min steady state light-moderate intensity cycling exercise (full text)
Day 2: 30 min cycle at an intensity ~65% VO2max, then 8 × 5 min intervals at 80% PPO, separated by 1 min active recovery; glycogen decreased by −266.9 ± 79.4 mmol/kg DM (−71%) (full text)
Biopsies from the vastus lateralis using the Weil-Blakesley conchotome technique, 2 h post-breakfast day 5 in each condition, breakfast providing 2 g/kg CHO (full text)
Food containers were weighed pre and post provision to account for leftovers (97.1% ± 1.0% compliance); photos of each meal were sent via online messenger (full text)
No significant difference following 4 versus 6 (p = 0.20) or 6 versus 8 g/kg/day (p = 1.00); increases of 107.1 ± 123.2, 35.3 ± 118.8 and 117.8 ± 137.3 mmol/kg DM; p = 0.47 (full text)
Stomach fullness median 1 (0–1), 0 (0–2) and 2 (0–4); Friedman p = 0.006; one female participant accounted for 55% of scores ≥ 5 (full text)
A 4:1 water:glycogen ratio would have increased body mass by 0.12, 0.70 and 0.58 kg, which was not detected (full text)
Pearson's correlation does not account for repeated observations; body mass/water measurements were possibly influenced by study procedures (full text)
Funded by Liverpool John Moores University Vice Chancellor PhD Scholarship awarded to R.O.J.; the authors declare no conflicts of interest (full text)