Are Carbs Really Driving Muscle Growth?

In one study of nine young men, adding 50 grams of maltodextrin to a 25-gram whey shake after four sets of single-leg extensions to failure made the insulin response, measured as area under the curve, about five times larger than with whey alone. Muscle protein synthesis did not follow: neither synthesis nor breakdown differed significantly between the two drinks, in the exercised or the rested leg ^[1]. The authors’ reading: once protein is dosed to stimulate synthesis maximally, extra insulin is not additive.

A later trial at rest found much the same when 60 grams of carbohydrate were added to 20 grams of protein ^[2]. But single-meal snapshots in small groups are not muscle gained over months. The question that matters for size is the long-term one: does a higher-carbohydrate diet, with protein held equal, build more muscle? In February 2026, a systematic review and meta-analysis in Sports Medicine pooled the randomized trials on exactly that ^[3]. Its answer is less tidy than either side of the debate would like.

The question behind the analysis

The case for carbohydrate as a growth nutrient usually runs through training: better acute performance, more volume over time, and insulin as a possible anabolic signal ^[3]. The International Society of Sports Nutrition’s 2017 position stand on nutrient timing reflects that view, stating that carbohydrate taken during resistance exercise, alone or with protein, supports greater acute and long-term training adaptations ^[4].

Menno Henselmans, Fredrik Tonstad Vårvik and Mikel Izquierdo asked a narrower question: does more carbohydrate change resistance-training hypertrophy when protein intake is the same in both arms ^[3]? Trials were excluded if protein intake differed between groups, whether by design or by a statistically significant margin, or if any supplement other than carbohydrate differed. That exclusion is what lets the authors speak of carbohydrate’s effect independent of protein; independence from total energy, which they also call for, is another matter. Two of the authors also co-wrote a 2022 review on carbohydrate and strength performance ^[5], and, given the mostly null earlier literature, the team hypothesized no significant effect ^[3]. That is transparent, and a reason to check that the methods, not the expectation, carried the conclusion.

How the review was built

The search covered MEDLINE, SPORTDiscus, SciELO and Google Scholar and was last updated in June 2025. Eligible trials were randomized, enrolled metabolically healthy adults aged 18 to 65, included at least six weeks of resistance training, and had arms that differed in carbohydrate by any amount, through diet or supplements. Of 1,347 records, 11 trials qualified ^[3].

They were small and short: about 21 participants and 8.5 weeks on average, with none longer than 12 weeks. Most trials enrolled trained participants; seven enrolled only men, three were mixed and one enrolled only women. The comparisons were lopsided, too. Eight trials pitted a ketogenic or very-low-carbohydrate diet against a higher-carbohydrate one, two tested carbohydrate supplements around training, and one compared a high-carbohydrate snack with a peanut-based one ^[3]. The evidence mostly describes the extreme end of restriction, not the gap between a moderate and a high intake.

Then there is what “hypertrophy” meant. Only two trials imaged muscle directly, with ultrasound. The other nine relied on whole-body fat-free mass from DXA, bioelectrical impedance or air-displacement plethysmography ^[3], a measure that also includes glycogen and the water stored with it.

Study quality on the TESTEX exercise-trial scale was fair to good, averaging 9.8 out of 15. Cochrane’s RoB2 tool found some concerns for most trials and a high risk of bias for one, and GRADE certainty of evidence was rated low ^[3]. That rating describes confidence in the pooled estimate, downgraded for imprecision and risk of bias; it is not the same as calling the trials poor. The authors also disclose that RoB2 and GRADE were completed during peer review, and that the review was registered on PROSPERO after its methods were set but before screening ^[3].

What the pooled data show

Across all 11 trials, the pooled difference in hypertrophy outcomes favored higher carbohydrate but was not statistically significant: SMD 0.15 (95% CI −0.10 to 0.40; p = 0.23), just under the 0.16 the authors treat as a small effect ^[3]. No individual trial was significant on its own, and the interval runs from a slight disadvantage to a benefit somewhere between small and moderate.

Heterogeneity was negligible (I² = 0), and removing trials one at a time left every estimate between 0.08 and 0.21 and non-significant. Dropping the trial with the largest positive effect, a 2018 crossover study in powerlifters and Olympic weightlifters, gave 0.08 ^[3]. Egger’s test and a visual check of the funnel plot showed no sign of publication bias, though with 11 small trials such checks have little power. The prediction interval, −0.19 to 0.49, is the more instructive number: the true effect in a new trial of this kind could plausibly be anything from a small disadvantage to a roughly moderate benefit ^[3].

An equivalence test, run post hoc in the paper’s discussion, says more about what the null result means. Its 90% confidence interval, −0.06 to 0.36, fell inside the bounds of a moderate effect but not inside those of a small one. Put plainly, the data argue against a moderate or larger hypertrophy benefit from higher carbohydrate under the conditions studied, but they cannot rule out a small one. The authors describe the results as most consistent with either a null effect or a small positive effect ^[3].

Energy, water and the limits of lean mass

Protein was roughly equal; energy often wasn’t

With protein differences excluded, the result speaks to carbohydrate at roughly equal protein, and much less to carbohydrate at equal energy. Even protein equality is approximate: in several trials participants ate freely and intake was self-reported, and in small samples a non-significant difference is not proof of a match ^[3].

Energy was looser still. Only 3 of the 11 trials met the authors’ isocaloric criteria, and together they reproduced the overall estimate with a far wider interval ^[3]. Ketogenic diets, the authors note, tend to suppress spontaneous food intake, and they flag three trials whose body-composition changes did not square with the intakes participants reported. Even the 2018 crossover study, classed as isocaloric, showed gains in fat and lean mass on higher carbohydrate and losses of both on the ketogenic diet; the authors estimate that this implies 5,337 kcal more positive energy balance on higher carbohydrate over the study period ^[3]. Without it, the two remaining isocaloric trials gave −0.16, which the authors treat as hypothesis-generating only.

A post hoc fat-mass analysis of 10 trials (one did not report fat mass) showed a non-significant trend toward more fat loss on lower carbohydrate (SMD −0.23; 95% CI −0.49 to 0.03; p = 0.09), and the authors argue that weak energy control may have exaggerated the apparent advantage of higher-carbohydrate diets ^[3]. With 8 of the 11 trials outside the isocaloric criteria, and the authors suspecting that the higher-carbohydrate arms were, on average, eating more, the slight lean toward higher carbohydrate cannot be confidently credited to carbohydrate itself. The same applies to claims that some people respond better to carbohydrate: the study the authors cite was non-randomized, and its higher-carbohydrate group also ate significantly more calories and protein ^[3]. That is a correlation with obvious confounders, not a demonstrated effect.

Lean mass is not the same as muscle

Fat-free mass also moves with glycogen and water. In the trial by Wilson and colleagues, participants who reintroduced carbohydrate in week 11, after 10 weeks on a ketogenic diet, gained a further 4.8% in fat-free mass, which the meta-analysis authors consider likely due to glycogen repletion ^[3]. Excluding that week barely moved the pooled result, from 0.15 to 0.175, so the refeed week did not drive the overall estimate. It does illustrate how a whole-body scan can register carbohydrate-driven shifts that are probably not contractile tissue. The two ultrasound trials, the only direct measures, leaned toward lower carbohydrate (SMD −0.26), but two trials cannot support a conclusion in either direction ^[3].

Performance is a separate question

None of this means carbohydrate does nothing in the gym; its best-supported benefit is acute. A 2022 systematic review of 21 crossover studies (226 participants) pooled 12 of them for total session volume: carbohydrate, compared with a placebo or water, increased it (SMD 0.61; 95% CI 0.11 to 1.11), with high heterogeneity and low certainty ^[6]. The effect was statistically significant in sessions longer than 45 minutes (SMD 1.02) and after fasts of eight hours or more (SMD 0.39). In shorter sessions it was smaller and not significant (SMD 0.23; 95% CI −0.21 to 0.67); after shorter fasts it was larger but imprecise and not significant (SMD 0.76; 95% CI −0.19 to 1.71). Maximal-effort sets were a significant moderator; carbohydrate dose was not, though its estimate was very imprecise ^[6].

A 2022 systematic review of 49 studies by Henselmans and colleagues was more conditional still. Acute benefits appeared in 6 of 19 studies, mainly those with fasted control groups and workouts above 10 sets per muscle group, and 15 of 17 long-term studies found no significant difference in performance changes with carbohydrate intake ^[5].

More fuel, more volume, more muscle: that is the usual rationale for carbohydrate-for-size advice. It is plausible, but the long-term trials did not confirm it. The 2026 authors note that sessions of 10 or fewer sets per muscle group generally reduce glycogen by about 40% or less, though type II fibers can be depleted further at higher volumes, and they argue that mechanical tension, not fuel availability, is the primary driver of hypertrophy ^[3]. The meta-analysis does not test whether the picture changes for very long, very high-volume or fasted training.

A 2022 meta-analysis of ketogenic diets in resistance-trained people (five studies, with fat-free mass pooled from four) also found no significant difference in fat-free mass, while flagging satiety and adherence as drawbacks ^[7]; several of its trials reappear in the 2026 analysis, so the two are not independent. The new paper questions carbohydrate as a direct driver of growth, not its role in glycogen restoration or endurance work: high-carbohydrate diets maximize glycogen stores, and high-volume exercise depletes them most ^[4].

What this means in practice

So are carbs driving muscle growth? Not in any way that 11 small trials with roughly equal protein could detect, and probably not by a margin the equivalence test would call moderate. That is a narrower statement than “carbs don’t matter,” and it rests on low-certainty evidence. The trials that could settle it — controlled feeding, direct muscle imaging, more than 12 weeks — are the ones the authors call for next ^[3]. Until then, carbohydrate is best treated as fuel and a practical route to an energy target, not as a demonstrated growth signal in its own right.

This is general information for healthy trained adults; anyone managing diabetes or another medical condition should discuss carbohydrate changes with a clinician.