REF-774 · Sherman et al., 1981 · International Journal of Sports Medicine, Vol. · Paper · PubMed

Effect of Exercise-Diet Manipulation on Muscle Glycogen and Its Subsequent Utilization During Performance

Blimited evidenceMeasured on trained athletesWritten before October 2026, being re-checked

Trained runners went through three six-day diet-exercise regimens before a 20.9-km performance run. A classic depletion-then-load diet (3 days at 15% carbohydrate, then 3 days at 70%) and a moderate version (3 days at 50%, then 3 days at 70%) raised gastrocnemius glycogen to 207 and 203 mmol/kg wet tissue, against 159 with 6 days at 50%.

The higher starting stores were used faster during the run (5.0 and 5.1 vs 3.1 mmol/kg per km), but run times and end-of-run glycogen did not differ.

The authors conclude that glycogen can be raised to high levels without the low-carbohydrate phase, that starting glycogen sets how much is used, and that carbohydrate loading brought no performance benefit for trained runners over 20.9 km.

Participants

Trained runners. The number, sex and age are not given in the abstract.

Protocol

Diets: 15% carbohydrate (L), 50% (M) or 70% (H) of intake.
Trial A: 3 days L, then 3 days H (classic depletion-load).
Trial B: 3 days M, then 3 days H (moderate).
Trial C: 6 days M (no loading).
In every trial a 5-day depletion-taper exercise sequence on the treadmill at 73% VO2max: runs of 90, 40, 40, 20 and 20 min. A day of rest, then the 20.9-km performance run.
Gastrocnemius biopsies on days 4 and 7 (on day 7 before and after the performance run).

Key Results

Trial A (15% → 70%)Trial B (50% → 70%)Trial C (50% throughout)
Muscle glycogen before the run (mmol glucosyl units/kg wet tissue)207203159
Glycogen used during the 20.9-km run (per km)5.05.13.1
Run timeno differenceno differenceno difference
Glycogen after the runno differenceno differenceno difference
Glycogen use in A and B was significantly higher than in C.
The abstract gives no run times and no p-values.

Practical Application

In trained runners given three six-day regimens before a 20.9-km run, a moderate taper with 3 days at 50% then 3 days at 70% carbohydrate raised muscle glycogen as far as the classic depletion diet (203 vs 207 mmol/kg wet tissue, against 159 without loading), but neither loaded regimen made the run faster: the extra glycogen was simply burned faster (5.0-5.1 vs 3.1 mmol/kg per km). You do not need the low-carbohydrate depletion days to load; and for a race of about half-marathon length, loading did not improve performance in this study.
The depletion phase is unnecessary on this evidence: the moderate regimen (50% then 70%) matched the classic 15%-then-70% regimen on stored glycogen.
Inference, not tested here: the 20.9-km run was probably not long enough for glycogen to run out in trial C (end-of-run glycogen did not differ), so the lack of benefit cannot be carried over to a full marathon, where running out of glycogen is the usual explanation for "the wall" (REF-775).
Not from this paper; the base's own practice: for a full marathon or longer, a two-to-three-day high-carbohydrate phase with a lighter training taper is reasonable, without a depletion phase; the base holds no marathon-length loading trial, so say that part is practice, not this study's finding.
Higher starting glycogen raised the rate of glycogen use; inference, not tested here: a loaded runner who starts too fast can spend the extra store early, so loading is not a licence for a faster opening pace.

Limitations

The number of runners is not given in the abstract; the design is small by the standards of its time and has no placebo control.
One performance distance (20.9 km); a longer run, where glycogen could run out, was not tested.
Diets set as a percentage of intake; grams per kilogram of body mass are not given in the abstract.
The abstract gives no run times, so the size of any non-significant difference is unknown.
Written from the abstract and record metadata only (full text not read).
How this entry was checked and changed
Why this level

three-trial comparison with an unreported n; one 20.9-km performance run per trial