The single most revealing sentence in the branched-chain amino acid literature is not in any paper. It is on the packaging, where BCAA is sold on the promise of muscle growth and the meta-analyses are about something else entirely.
Four recent meta-analyses tell the story cleanly, and it goes like this: pooled across trials, BCAA plus resistance training moves one body composition index by a standardized mean difference of 0.337, and moves nothing else measurable. A separate meta-analysis finds improvements in strength and physical function in older adults — and then rates the certainty of that evidence as low to very low by its own GRADE assessment. And the largest and most consistent effect in the whole field is not on muscle at all. It is on how sore you are two days after an eccentric session.
That is a real effect. It is also not what the tub says on it.
The muscle meta-analysis, read carefully
The 2026 meta-analysis in Archives of Gerontology and Geriatrics pooled twelve randomized trials with 459 older adults with sarcopenia, all of whom were doing resistance training, half of them with a BCAA-rich supplement added.
The result that reached statistical significance: skeletal muscle index, standardized mean difference 0.337, 95% confidence interval 0.035 to 0.639, p = 0.029.
Everything else was null. Grip strength SMD 0.289, p = 0.080. Thirty-second chair stand, p = 0.169. Short Physical Performance Battery, p = 0.124. Walking speed, p = 0.283. Knee extension strength, p = 0.328. Total lean mass, p = 0.908 — which is to say, no difference whatsoever. Fat mass, p = 0.866.
Two things are worth noting. The first is the interval: 0.035 to 0.639 on a scale where zero is nothing, means the trial small enough to plausibly be the upper end. The second is that total lean mass — the simplest, most direct measure of “did people grow” — came out at essentially zero difference, while a ratio-based index came out at 0.337. A skeletal muscle index is lean mass scaled by height or by body size, and a ratio can move when lean mass does not, if fat mass also moved. The authors’ own conclusion is appropriately narrow: BCAA after resistance training increased the skeletal muscle index in older adults, without improving other functional or body composition outcomes.
That is a fair, careful sentence, and it is a long way from a muscle-building claim.
The strength meta-analysis, and its own grade
The 2026 systematic review in BMC Musculoskeletal Disorders is the more interesting of the two, because it reports both the effect and how much to trust it.
Nine randomized controlled trials, 496 participants with sarcopenia, resistance training plus amino acid-based supplementation versus resistance training alone. And here the findings are more positive: handgrip strength SMD 0.69 (0.04 to 1.35), gait speed SMD 0.64 (0.02 to 1.25), Short Physical Performance Battery SMD 1.69 (0.81 to 2.57), five-times sit-to-stand SMD −1.42 (−2.68 to −0.16). All statistically significant.
Now read the framing sentence, because it changes how you read the numbers: the evidence is rated low to very low certainty according to the GRADE framework used in the paper. And the effect on muscle mass index and appendicular lean mass was not significant.
So: a large, statistically significant effect on function, no effect on muscle mass, and low certainty. That combination has a specific explanation, and the paper offers it — subgroup analysis found that baseline protein intake, intervention setting, supplement type, training frequency and whether the supplement contained co-nutrients were all significant effect modifiers. When you find that the answer depends on whether the supplement had other things in it, you are looking at a supplement literature rather than an amino acid literature.
A 0.69 effect on grip strength is a large effect. A large effect at low certainty is usually a sign that the pooled estimate is doing more work than the individual trials support.
The strongest effect in the field is about soreness
The 2024 meta-analysis in Sports Medicine – Open is the best-designed of the group: eighteen studies, thirteen of them rated high quality, pooled with Hedges’ g and meta-regression on the modifiers.
The endpoint is exercise-induced muscle damage, and the results are unambiguous. Creatine kinase fell immediately after damaging exercise at g = −0.44 (p = 0.006) and at 72 hours at g = −0.99 (p = 0.002), with no significant effect at 24, 48 or 96 hours. Delayed onset muscle soreness fell at 24 hours at g = −1.34, at 48 hours at g = −1.75, at 72 hours at g = −1.82, and at 96 hours at g = −0.82, all significant, with no effect immediately after. Lactate dehydrogenase did not move at any time point.
Meta-regression found that higher daily and total amounts of BCAA, and longer periods of use before the damaging session, were associated with the largest effects on creatine kinase and on soreness.
Soreness effect sizes of −1.34 to −1.82 are not small. In exercise research those are large standardized effects, and they are the kind of number that changes what a person experiences in the week after a hard session. The question is what that experience is worth, and the answer depends on whether soreness is a cost you were trying to avoid or a signal you were using to judge whether the session worked. The meta-analysis does not answer that, because nothing in it measures long-term adaptation.
The acute mechanistic story, and the hole in it
The case for BCAA starts with leucine, which among the three has been studied most because of its ability to trigger the initial acute anabolic response. Leucine, isoleucine and valine together make up about a third of muscle protein. A 2021 narrative review in the International Journal of Sports Nutrition and Exercise Metabolism does the useful work of describing where the support actually comes from, and it is worth quoting the authors’ own framing: early support for BCAA supplementation was derived from extrapolation of mechanistic data on their role in muscle protein metabolism.
That is the honest origin story. The hypothesis is: leucine triggers synthesis, therefore supplement leucine, therefore you grow. The review’s purpose is to scrutinize that chain against the acute and longitudinal research, and the field has not closed the gap.
The 2023 quantitative systematic review in Physiology Reports tested the specific mechanism directly. It pooled studies in healthy adults who did a bout of resistance exercise, ingested a protein bolus within an hour, and had both plasma leucine and muscle protein synthesis measured. Ingested leucine was associated with the magnitude of the synthesis response in older adults over the acute period (r² = 0.64, p = 0.02) and over the whole postprandial period (r² = 0.18, p = 0.01) — but not in younger adults. And here is the finding that undercuts the threshold story: the authors identified no threshold in older adults, and no single plasma leucine variable — peak, rate of rise, total availability — had substantial predictive capacity over the response in either age group.
So the mechanism the products are sold on is real in the laboratory, and does not resolve into a usable threshold in humans.
Why the acute mechanism keeps producing null long-term results
The trial with 30 postmenopausal women, published in Experimental Gerontology in 2021, is the cleanest illustration. Three arms: resistance training plus placebo, resistance training plus nine grams a day of BCAA, and a no-training control. Eight weeks of supervised whole-body training, three sessions a week, three to four sets at 60 to 75% of one-repetition maximum.
Muscle mass and strength increased significantly in both training arms. Myostatin fell and IGF-1 rose in both. Follistatin rose significantly only in the BCAA arm — a real change in a real regulatory marker. And there were no differences between the two training conditions over time. The no-training control changed in nothing.
The authors’ conclusion: eight weeks of resistance training is effective on its own, and adding BCAA failed to augment the change.
The reason is not mysterious, and it is stated in the meta-analysis on protein and resistance training from the British Journal of Sports Medicine in 2018. Across 49 studies and 1,863 participants, protein supplementation added 2.49 kg to one-repetition maximum strength and 0.30 kg to fat-free mass, and the incremental benefit fell with increasing age and rose with training experience. Above 1.62 grams of protein per kilogram of body weight a day, there was no further gain in fat-free mass from adding more.
Supplements do not create an environment where protein was previously limiting. If you are eating enough protein, adding isolated amino acids adds nothing to a total that was already sufficient. That is not a defect of the BCAA studies — it is the reason the null results cluster in people who were already well fed.
What these meta-analyses do not show
None of them has a muscle growth endpoint in trained young adults. The muscle-mass data come from older adults with sarcopenia, where the intervention target is the arrest of loss, not the addition of muscle. The growth story and the preservation story are different claims and the field has evidence only for the second.
Not one of these meta-analyses has measured a hard outcome. No strength-to-failure injury rate, no falls, no hospital admission, no functional independence, no mortality. Everything here is a surrogate: an index, a ratio, a standardized mean difference of a hand dynamometer reading.
The certainty rating on the most positive meta-analysis is low to very low, by the authors’ own assessment. That is not a caveat at the end of the paper, it is the paper’s own conclusion about its own findings.
Total lean mass did not move in either meta-analysis. If the mechanism is synthesis, the most direct downstream measure should respond. It did not, in the BCAA pool (p = 0.908) and not in the broader amino acid pool (no significant effect on muscle mass index).
The soreness finding has never been connected to an adaptation finding. Large effects on a 24-to-96-hour recovery marker in eighteen studies, and not one trial in these meta-analyses asks whether the people who hurt less grew more.
The comparator is almost always placebo, not protein. Isolated BCAA against a placebo capsule, in people whose total daily protein is not reported. The intervention is only meaningful if protein is limiting, and almost none of these trials establishes that it was.
The populations are narrow. Postmenopausal women in one trial, older adults with sarcopenia in the two meta-analyses. A man of 30 training four times a week is not in this literature.
The funding picture is not reported in any of these abstracts, which is a hole in its own right. Nobody reading the pooled estimates can tell which of the twelve or the nine or the eighteen trials were paid for by a seller.
The honest version, then: BCAA is a real product with a real effect on one real thing — how much you hurt in the days after damaging exercise — and a plausible mechanism that does not translate into measurable growth when the person eating the powder is already eating enough protein. The comparison that everyone actually needs, the one nobody has run, is BCAA against a matching amount of ordinary dietary protein, measured in people training hard, for a year.
