Deep divefrom Episode 10Training7 November 202615 min

Once, Twice or Three Times a Week for Muscle?

A deep dive into one paper: what was known before, what the researchers did, what the numbers say and what they don't.

The paper: Resistance Training Frequency and Muscle Hypertrophy (medRxiv preprint)

Chapters

Transcript

The script the two AI voices read from. The times are worked out from the script, not heard, so a line can start a few seconds early or late.

Read the whole episode

Cold open

Host 1You heard the trailer, and this is the whole story. Welcome to the Hastrup Review of Science, where artificial intelligence reads real science, so you don't have to pretend you understood the paper.

Host 2Today is one paper, a systematic review and meta-analysis that asks a gym question people have argued about for decades. If you do the same total amount of work for a muscle each week, does it matter whether you split it into one, two or three sessions?

Host 1So the question isn't whether to train more, it's whether to spread the same training out differently across the week?

Host 2Exactly that. The amount stays fixed, only the spacing changes, and the outcome is how much the muscle grows.

Why anyone cares

Host 1Before we get to the answer, why have people been fighting about this for so long?

Host 2Because there are two old traditions in strength training that point in opposite directions. Bodybuilders have long favoured the so-called split, where each muscle gets one hard session a week, while many coaches and athletes train the whole body two or three times a week.

Host 1And each camp thinks its way builds more muscle?

Host 2Each camp has a story for why. The split camp says one big session gives the muscle a strong signal and plenty of time to recover, and the frequent camp points to a biological argument.

Host 1What is that biological argument, in plain words?

Host 2In the wider physiology literature, a hard session switches on muscle protein building for roughly a day or two, and then it fades back down. The idea is that if you only train a muscle once a week, it spends most of the week idle, so hitting it more often should keep the building process switched on more of the time.

Host 1That sounds pretty convincing, so why isn't it just settled?

Host 2Because a plausible mechanism is not the same as a measured result. Muscle growth over months depends on much more than a short-term signal, and the only way to know is to put people on different schedules and measure their muscles.

What was known before

Host 1What did the field already know before this paper came along?

Host 2The broad picture from earlier reviews is that the total weekly amount of work, usually counted as hard sets per muscle per week, is the main driver of growth. This paper's own abstract opens on that point, calling weekly volume the primary driver.

Host 1When you say volume, do you mean how heavy the weights are?

Host 2Not quite, it means how much work you do, typically the number of hard sets, or sets times repetitions times load. Heavy or light can both build muscle if the sets are taken close to failure, but more weekly sets generally means more growth, up to a point.

Host 1So where does frequency come in, if volume is doing most of the work?

Host 2That is the tangle. In many older studies, the group that trained more often also simply did more sets in total, so you couldn't tell whether frequency helped or whether it was just the extra volume.

Host 1So if I have this right, the people training three times a week might have grown more just because they did more work, not because of the schedule?

Host 2Yes, that is the confound. Earlier reviews that mixed both kinds of trial tended to find some edge for training a muscle at least twice a week, but when analysts looked only at trials that held volume equal, that edge largely shrank away.

The open question

Host 1Then what was still left open, if that had already been looked at?

Host 2Two things. First, how big any remaining frequency effect could plausibly be, with a proper range around it, and second, how confident we should be, given that the trials are small and measure muscle in a particular way.

Host 1Why does the range matter so much, if the answer is just yes or no?

Host 2Because no difference found is not the same as no difference exists. A good review tells you the biggest effect the data still allow, so you know whether a small advantage could be hiding in there.

What this paper did

Host 1So what exactly did these researchers do?

Host 2Two authors ran a systematic review and meta-analysis, following a reporting standard called PRISMA twenty twenty. They searched five research databases up to October twenty twenty six for training trials that met strict rules.

Host 1What does PRISMA actually guarantee, for someone who has never heard of it?

Host 2It is a checklist for how a review is reported: how the search was done, how studies were screened in or out, and why. It doesn't make a review correct, but it makes it possible for others to see and repeat each step.

Host 1What were the strict rules a trial had to meet to get in?

Host 2Three main ones. The trial had to last at least six weeks, it had to equate weekly volume between the groups training a muscle at different frequencies, and it had to measure muscle size.

Host 1Why six weeks, is that long enough to see a muscle grow?

Host 2It is about the minimum, and the wider training literature suggests measurable growth usually shows up after several weeks of consistent work. Shorter trials risk picking up swelling from recent sessions rather than real new tissue.

Host 1And the frequencies they compared went from what to what?

Host 2From one to three sessions per muscle per week. So one versus two, two versus three, and one versus three, all with the same total work.

Host 1How many studies and people did they end up with?

Host 2Sixteen trials with three hundred and eighty participants in total. That works out to roughly two dozen people per trial on average, which is small for each trial on its own.

Host 1How do you combine sixteen different studies that might have measured things differently?

Host 2They turned each trial into one standard number called Hedges g. It takes the difference in muscle growth between the two schedules and divides it by how much people naturally vary, so studies using different measuring units can sit on the same scale.

Host 1Can you give me a feel for what a Hedges g of a certain size means?

Host 2By the usual rule of thumb, around zero point two is a small effect, zero point five medium, and zero point eight large. Anything close to zero is called trivial.

Host 1You said one number per trial, but many studies measure several muscles, so how did they deal with that?

Host 2They kept to one change-score effect per trial, meaning the growth from start to finish in one group compared with the other. That avoids letting a trial with many measurements count several times and swamp the others.

Host 1And which direction is good news for training more often?

Host 2In this paper, a negative number favours the higher frequency. So a number below zero would mean the more frequent schedule grew more muscle.

Host 1Then they just averaged the sixteen numbers together?

Host 2A weighted average, using what is called a random-effects model. It assumes the true effect may genuinely differ a bit from study to study, because of different people, programmes and muscles, and it weights larger, more precise trials more heavily.

Host 1Is there any way they rated how much to trust the final answer?

Host 2Yes, they used a system called GRADE. It starts from the type of evidence and then marks it down for problems like small studies, risk of bias, inconsistency or indirect measurement, ending in a rating from high to very low.

A word from our sponsor

Host 1This episode is brought to you by Spare Day, a limited edition eighth day of the week, poured by hand into a small glass jar.

Host 2I keep mine in the fridge door, next to the mustard.

Host 1Open the jar on any Sunday night and your week gets one extra day, so you can spread the same weekly volume across one more session.

Host 2I used mine for calf raises. Sixteen trials, three hundred and eighty participants, and no measurable difference whatsoever, which I find very restful.

Host 1Only a few hundred jars will be made, because after that the calendar runs out of weekdays to borrow from.

Host 2Mine has already been preregistered as a Thursday.

Results

Host 1So after all that pooling, what did they find?

Host 2Essentially nothing. The pooled effect was a Hedges g of minus zero point zero two, which is trivial and not statistically significant.

Host 1How close to zero is minus zero point zero two, on that scale you described?

Host 2Very close, about one tenth of what would count as a small effect. On its face, it means the more frequent schedules grew muscle very slightly more, by an amount nobody could notice.

Host 1How sure can they be that the real answer isn't hiding a bit higher or lower?

Host 2The ninety five percent confidence interval ran from minus zero point two one to plus zero point one six. So the data are compatible with anything from a small advantage for higher frequency to a slightly smaller advantage for lower frequency.

Host 1So if I have this right, the best case for training more often is a small benefit, and not more than that?

Host 2That is how the authors put it: the interval admits at most a small effect favouring the higher frequency. A medium or large advantage is not supported by these trials.

Host 1Did the sixteen studies disagree with each other, or did they all point the same way?

Host 2The measure of disagreement, called I squared, came out at zero percent. That means the spread between trials was no bigger than you would expect from chance alone.

Host 1Does zero percent mean the studies were all perfectly consistent?

Host 2Not necessarily, and that's worth flagging. With small trials and wide uncertainty in each, real differences between studies can be hard to detect, so zero partly reflects how little power there is to see them.

Host 1What about the separate comparisons, like once a week against twice a week?

Host 2The contrast-specific estimates were similarly trivial. No particular pairing, one versus two, two versus three or one versus three, broke away from the overall picture.

Host 1Did they check whether it matters for certain people, like beginners versus experienced lifters?

Host 2They ran moderator analyses, which test whether the effect changes with features of the study or the people. None was significant, but the authors are candid that those analyses could only have detected large effects.

Host 1You also mentioned a prediction interval, how is that different from the confidence interval?

Host 2The confidence interval is about the average effect, while the prediction interval estimates the range you would expect in a new, future trial. Here it ran from minus zero point two three to plus zero point one eight, so even a fresh study would be expected to land in trivial to small territory.

Host 1And how much did GRADE say we should trust all this?

Host 2Low to very low certainty. That is a sober rating, and it means future research could quite plausibly shift the estimate.

Why it matters

Host 1Why does a small difference in muscle growth even matter to anyone?

Host 2The principle here is that, within the range tested, how you schedule a fixed amount of work seems to matter far less than the amount itself. The authors conclude the data give no basis for preferring one, two or three sessions per muscle at a fixed weekly volume.

Host 1So for an ordinary person who goes to the gym, what does that change?

Host 2It suggests you can pick the schedule that fits your life. If two longer sessions suit your week better than three short ones, these trials give no reason to think you will grow noticeably less, as long as the total work is the same.

Host 1Is that the same as saying frequency never matters at all?

Host 2No, and it's an important distinction. In practice, training more often is one of the easiest ways to fit in more total sets, and quality may drop in a single very long session, so frequency can matter indirectly through volume, which this paper deliberately held fixed.

Host 1Who would this matter for beyond regular gym-goers?

Host 2Coaches writing programmes and people with tight schedules, for whom fewer sessions is the only realistic option. It also matters to researchers, because it tells them where the remaining uncertainty sits.

Host 1What can't we conclude from this paper, even if the numbers look clean?

Host 2We cannot conclude anything about more than three sessions per muscle per week, because that is outside what was tested. And we cannot rule out a small advantage, or say anything firm about subgroups like older adults or very advanced lifters.

Honest caveats

Host 1What are the biggest weaknesses the authors themselves point to?

Host 2The most important one is about measurement. Most trials measured muscle thickness, and the authors note that it does not track whole-muscle growth reliably.

Host 1How do you actually measure muscle thickness, and why would it miss growth?

Host 2Usually with ultrasound, which images the depth of the muscle at one or a few spots on the limb. Muscles can grow unevenly along their length, so one spot may miss growth elsewhere, whereas whole-muscle volume generally needs an MRI scan, which is costlier.

Host 1So the study might be measuring the right muscle in the wrong place?

Host 2That is a fair way to put the risk, though it isn't a sign of bias toward one schedule. It adds noise and indirectness, which is partly why GRADE ended up low.

Host 1Are there other limits, besides the measuring?

Host 2Small trials, about two dozen people each, and only sixteen of them. With three hundred and eighty people in total, a genuinely small benefit could slip past unnoticed, which is exactly what the confidence interval is warning about.

Host 1Is there anything about the timing of the studies that could hide an effect?

Host 2Most were relatively short, with six weeks as the minimum, so we don't know whether small differences might add up over a year or more. That is not tested here.

Host 1Is this paper published in a journal yet?

Host 2It is a preprint on medRxiv, which means it has not yet been through formal peer review. The methods follow the standard playbook, but outside reviewers haven't yet checked it.

What to watch next

Host 1What would you want to see next, to settle it for good?

Host 2Larger trials that measure whole-muscle size, ideally with MRI, and that run for many months. That would shrink the uncertainty and show whether a small benefit is real.

Host 1Would you also want studies that push past three sessions a week?

Host 2Yes, frequencies above three per muscle are much less studied, and they matter for athletes who train daily. Trials in specific groups, such as older adults trying to keep muscle, would also help.

Host 1And if a big new trial does find a small edge for training more often, would that overturn this paper?

Host 2Not really, it would fit inside what this paper already allows. The interval explicitly leaves room for a small benefit, so such a finding would refine the picture rather than reverse it.

Sign-off

Host 1So if I had to sum it up for a friend, it's: do the work, and spread it however suits you?

Host 2Within one to three sessions a week, at equal volume, that is a fair summary of what this evidence supports, with low certainty and a small effect still possible. Thanks for listening to Hastrup Review of Science.

Host 1Every weekend this same feed carries the roundup, the week's new papers in one short episode, which is roughly as often as anyone needs to hear from us.