The bathroom scale is the wrong instrument for any of this.
That sentence is the simplest version of what the recent body composition literature on GLP-1 receptor agonists is asking the public conversation to take seriously. The conversation has settled into a recognisable groove: the medications work, but you lose muscle, and that’s the catch. There’s a piece of truth in the framing. There’s also a meaningful distortion. The reality the imaging studies of the last five years describe is more nuanced than the public summary, and the practical implications for anyone considering or already on these medications are different from the standard advice.
Three things have to be held together to think about this clearly. The body composition shift on GLP-1s is real but proportional: the same ratios you see with any sustained weight reduction. The fat that’s being lost is preferentially the dangerous kind, not the harmless kind, and that distinction matters more than total weight loss does. The measurement landscape (what you can actually find out about your own body composition with the tools available to you) is much weaker than the conversation usually acknowledges.
What “muscle loss” actually means in the data
Lean mass loss happens. The headline finding everyone has heard is real. The 2024 network meta-analysis pooling 22 randomised trials with 2,258 participants found that GLP-1 receptor agonists produce, on average, a lean mass reduction that accounts for approximately 25 percent of total weight loss. That’s the figure that has driven the public concern.
The figure is correct. In context, it’s also unremarkable. The same 25-percent ratio is what the body composition literature has reported for sustained weight reduction by almost any method: bariatric surgery, very low calorie diet, slow caloric restriction over years. The body, when it sheds weight, sheds some lean tissue as part of the package. The ratio is biologically determined, not pharmacologically determined, and the GLP-1 pathway doesn’t appear to disturb it.
The corollary, which has received less coverage, is that the relative proportion of lean mass to total body weight stays essentially unchanged. The STEP 1 substudy is the largest dual-energy X-ray absorptiometry assessment of semaglutide-induced body composition change. Published in 2021, it tracked 140 participants over 68 weeks. The headline weight reduction was 15 percent of body mass on semaglutide versus 3.6 percent on placebo. Total fat mass dropped by 19.3 percent. The proportion of body weight that was lean tissue, however, stayed roughly stable across the trial. The body shrank. The composition stayed in the same neighbourhood.
That’s what the data is actually saying. It’s not saying that GLP-1s are uniquely catabolic. It’s saying that any meaningful weight loss takes lean tissue with it, in roughly the same proportion, and that the question worth asking is not whether to prevent the lean loss but whether to modulate it.
Where the fat actually goes
Here’s what’s interesting in the body composition data, and what the public conversation has largely missed.
In the same STEP 1 DEXA substudy, total fat mass dropped by 19.3 percent. Visceral fat, the deep abdominal fat that wraps around organs and drives the metabolic disease that obesity is mostly dangerous for, dropped by 27.4 percent. The visceral component reduces faster than total fat does. The fat that matters most for cardiovascular and metabolic risk is preferentially mobilised. The fat that’s largely cosmetic and metabolically inert, the subcutaneous layer, is more resistant to loss, as you’d expect, because it’s always been the body’s preferred long-term energy store.
The story extends to ectopic fat, the fat that infiltrates organs themselves and particularly the liver. The 2021 randomised trial of semaglutide in non-alcoholic fatty liver disease, using MRI proton density fat fraction as the measurement, found liver fat reductions of approximately 30 to 50 percent versus placebo over 24 to 72 weeks. That’s the most metabolically dangerous fat in the human body, and the medication pulls it out preferentially.
This is the part of the story the public conversation has largely missed. The “you lose 25 percent muscle” framing treats the body as homogeneous, as if a kilogram of fat is a kilogram of fat, and a kilogram of muscle is a kilogram of muscle, and the only number that matters is the total. The clinical reality is different. The kilogram of visceral fat lost from around the liver and the kilogram of subcutaneous fat lost from the thighs aren’t equivalent in any meaningful sense. The first reduces lifetime risk of cardiovascular disease, type 2 diabetes, fatty liver disease, and several cancers. The second changes how the trousers fit. GLP-1s, the data suggests, tilt the loss toward the first category.
If the right unit of analysis is what this body composition shift means for ten-year health outcomes, rather than what’s happening on the scale, the picture becomes considerably more reassuring than the muscle-loss panic suggests.
Why marathon-runner calves are the right intuition
There’s a second confusion worth being explicit about, because it shows up everywhere in the public discussion.
Watch the calves of an elite marathon runner. They aren’t large. By any visual measure they’re slimmer than the calves of someone of comparable height who does no specific training at all. They’re also, by every measure of function that matters, repeated contraction over hours, oxygen uptake per gram of muscle, mitochondrial density, capillary supply and fatigue resistance, outperforming the larger untrained calf by orders of magnitude. The bulk doesn’t translate to capacity. The bulk is the wrong outcome.
The body composition literature on GLP-1s reads similarly. The most recent direct evidence is a paper published in Cell Reports Medicine in 2026 from researchers at the Helmholtz Munich diabetes centre. They examined GLP-1-induced weight loss in mice and humans and asked the question most of the prior literature hadn’t asked precisely: does the absolute reduction in muscle mass translate to a reduction in muscle function? Their answer was that it doesn’t, or not in any clinically detectable way. Strength tests, walking speed, sit-to-stand performance, mobility: the functional outcomes were preserved or improved despite the gross muscle mass reduction.
This is consistent with what we know about muscle quality more broadly. Muscle mass and muscle function are correlated but distinguishable. The factors that drive function, fibre type composition, neural recruitment patterns, intramuscular fat infiltration (myosteatosis) and capillary density, are responsive to training, not to total mass. A trained smaller muscle outperforms an untrained larger one on every functional axis. A scale, or even a DEXA scan, can’t distinguish between them.
The implication for someone on a GLP-1 is specific. The right outcome to track isn’t your total lean mass on a body composition scan. It’s your functional capacity. What you can lift. How far you can walk before fatigue. How easily you can rise from a chair. How stable you are on your feet. If those numbers are stable or improving, the gross mass change is largely cosmetic. If those numbers are degrading, the conversation about protein, training, and dose adjustment is the right conversation.
What you can actually measure
This is the harder part of the conversation. The measurement gap between what would be useful and what’s available is real, and pretending otherwise would be unfair.
The clinical gold standard for body composition is dual-energy X-ray absorptiometry, a DEXA scan. It separates fat mass, lean mass, and bone mass with reasonable precision, and most major Australian cities have private DEXA providers. The cost is typically $100 to $150 per scan, not Medicare-rebated for body composition assessment outside specific clinical indications. The radiation dose is small. For someone serious about tracking body composition through a sustained intervention, two scans, one at baseline and one at 12 months, are probably worth the cost.
The research gold standard is magnetic resonance imaging, including specialised techniques like MRI proton density fat fraction for liver fat and magnetic resonance spectroscopy for intramuscular fat. These are research tools rather than routine clinical ones. Most patients won’t access them. They’re worth mentioning here to make the point that the scientific literature describing what GLP-1s do to body composition is built on imaging that very few patients will personally encounter.
What most patients have at home is a bioimpedance scale, the kind that runs a small electrical current through the body and estimates body composition from the resistance. The honest assessment of these scales is that they’re imprecise but not useless. The absolute numbers, your fat mass on any given day, in any given hydration state, at any given time of day, shouldn’t be trusted to within several percentage points. The trends, taken on the same scale at the same time of day under similar conditions over weeks and months, are roughly directional. They’re better than the bathroom scale alone. They’re not a substitute for a DEXA.
What is genuinely useful, and almost free, are the functional and anthropometric measures the medical literature has been recommending for years. Waist circumference, measured consistently, correlates well with visceral fat. Grip strength, measured with a $50 dynamometer, is one of the strongest single predictors of overall functional capacity in adults. Sit-to-stand performance, counting how many times you can rise from a chair without using your hands in 30 seconds, tracks lower-body function. Walking pace and time to fatigue are simple to track. None of these require a clinic. All of them, tracked over months, give a more honest picture of what’s happening to your body than the scale or any consumer scan.
What actually preserves the things worth preserving
The body composition data points consistently in one direction on the modulating intervention. Resistance training isn’t a nice-to-have. It’s the variable that determines whether the lean tissue lost on a GLP-1 is the proportional, expected, biologically-determined fraction or something larger.
The most direct evidence is the Copenhagen S-LiTE trial published in The Lancet’s eClinicalMedicine in early 2024: a four-arm randomised study of post-weight-loss patients maintained for one year on either supervised exercise alone, GLP-1 alone, the combination, or placebo, then followed for a second year off treatment. The combination of exercise plus GLP-1 produced the best body composition outcomes during treatment. More importantly, it was the only arm that maintained both weight loss and lean mass a full year after all active treatment ended. Exercise alone partially maintained outcomes. GLP-1 alone produced regain. Placebo regained most.
The exercise component in S-LiTE was supervised resistance training plus moderate aerobic work, two to three sessions per week. This isn’t a punishing prescription. It’s the unglamorous middle ground that the long-term outcomes data has been pointing at for two decades. The protein intake recommendation that accompanies it, broadly 1.6 to 2.2 grams per kilogram of goal body weight distributed across meals, is well-supported in the broader sports nutrition literature and is consistent with what most clinicians experienced in this space recommend.
A second variable, less discussed, is the choice of medication itself. The 2024 network meta-analysis is quietly informative on this. Liraglutide, the older and weaker GLP-1 receptor agonist, produces less total weight loss but the smallest lean mass effect. Semaglutide and tirzepatide, the newer and more potent agents, produce more total weight loss but with proportionally more lean mass loss. There’s a real trade-off, and it’s a real clinical conversation. For some patients, particularly those already at the lower end of muscle mass for their age or with frailty risk, the slower path with the older agent may be the right path. For others, the faster total loss with appropriate exercise modulation will be the better trade. The choice shouldn’t be made on advertising volume.
A third consideration is dose pacing. The lean mass effect appears more pronounced with rapid weight loss than with slower reduction. There’s a clinical case for titrating doses upward more slowly than the manufacturer’s standard schedule, particularly during periods when resistance training intensity is being established or when injury or illness has interrupted exercise.
What this means in practice
For someone considering or already on a GLP-1 receptor agonist, the summary is this:
The lean mass shift the public conversation worries about is real but proportional, and unremarkable in the broader weight loss literature. The fat reduction is preferential: the visceral and ectopic fat that drives metabolic risk drops faster than the cosmetic subcutaneous layer. The functional outcomes, strength, mobility, walking speed and fatigue resistance, are largely preserved despite the gross mass change. The bathroom scale is a poor instrument and a body composition scan is a much better one, but the most honest measure of all is what your body can do. Resistance training isn’t optional, and the dose of training matters more than most patients are told. The drug choice is a conversation worth having with someone whose job is the long arc, not the prescription pad.
The standard advice, eat protein and exercise more, isn’t wrong. It’s incomplete. The reality is more nuanced than the standard advice supports, and knowing the nuance changes what to track, what to worry about, and what to actually do.
The body that walks out the other side of a year on a GLP-1, with appropriate clinical conversation and a serious approach to resistance training, is in most measurable senses a healthier body than the one that walked in. The visceral fat is lower. The liver fat is lower. The functional capacity is preserved. The cardiovascular risk profile is better. What the scale shows is the least interesting of these outcomes, and yet for most patients it is still the only number anyone is paying attention to.
That’s the conversation worth changing.