Diet · GLP-1s
A quarter of the weight lost is lean mass, and that is also true on placebo
The muscle loss argument about these drugs is usually conducted with the wrong number. The ratio of lean to fat in the weight people lose is unremarkable and the drug barely changes it. The amount is what changed, and the two things get reported as if they were one.
The Archive’s position
This is opinion, not a finding. Everything else on this page is the evidence.
The muscle and hair loss people get on a GLP-1 is mostly not the drug doing something to them. It is what happens when appetite disappears and somebody simply stops eating instead of eating differently. Hit your protein target and keep lifting, and most of it is avoidable.
Why we hold it
- The lean mass ratio is the same on the drug and on placebo. In the SURMOUNT-1 scans about 25% of lost weight was lean tissue in both arms, so the drug is not changing what weight loss is made of. It is changing how much of it there is.
- Rapid weight loss by any route does this. The same hair shedding pattern is well documented after bariatric surgery, on a much larger scale than the 3.9% seen with these drugs, which points at the speed of the loss and at what is being eaten rather than at the molecule.
- Two 2026 meta-analyses disagree about whether these drugs are worse than dieting for lean mass. They agree completely about what helps. Adding resistance training produced the lowest share of weight lost as lean tissue in both, 7.7% in one and 17.5% in the other, against roughly 22 to 26% for dieting alone. When two analyses that contradict each other point the same way on one variable, that variable is the one to act on.
- Appetite suppression removes the feedback that normally makes people eat. Nobody defends an 800 calorie day on purpose; it happens because nothing feels urgent. A protein target is a number you hit deliberately, which is exactly what is needed when hunger stops doing the reminding.
Where this goes beyond the evidence
No trial has tested this prescription. The one trial that added protein without training found nothing, and it is quoted everywhere as proof that protein does not help: it compared 52 with 77 grams a day on a 600 calorie diet, both around or below the RDA, so it tested a low intake against a slightly less low one and cannot speak to 150 or 200 grams. That cuts both ways. It does not refute the position, and it does not support it either. The trial that would settle it, randomising people on a GLP-1 to resistance training, to 1.6 g/kg of protein, to both, or to neither, is recruiting now and reports in 2029. The claim that this is mostly avoidable is a reasonable inference from the training data plus the mechanism. It is not a measured result, and the evidence is stronger for the training half of the advice than the protein half.
Who this is not for
Anyone reading this as a reason to stop or adjust a prescribed medication. Nothing here is about whether to take these drugs, which is a decision with a doctor. There is also a genuine gap underneath all of it: across 35 randomised trials, not one reported an objective measure of physical function, so what any of this lean mass change means for strength or for daily life has largely not been measured.
Two corrections that change the size of the problem
The first is the ratio against the amount. In the SURMOUNT-1 scans, about 75% of the weight lost was fat and 25% was lean, and that split was the same in the placebo group. So the drug is not altering the composition of weight loss. It is producing a great deal more of it, 21.3% of bodyweight against 5.3%, and a normal ratio applied to an unusual quantity still removes an unusual quantity of lean tissue. Both sides of this argument tend to pick whichever of those two facts suits them.
The second is that lean mass is not muscle. A DXA scan’s lean compartment includes organs, connective tissue, skin and body water, and skeletal muscle is roughly half of it. Every headline percentage in this discussion is therefore about double the figure for muscle itself. When thigh muscle was measured directly by MRI in SURPASS-3 and compared against longitudinal population data, the volume lost was broadly what the relationship between bodyweight and muscle would predict, and the fat infiltrating the muscle improved.
What survives both corrections is still worth acting on. Losing that much weight that fast takes real tissue with it, and the trials say clearly which lever moves that number most.
What was measured
About 25% of the weight lost was lean mass, on the drug and on placebo alike
Measured in
160 participants in the SURMOUNT-1 DXA substudy, scanned at baseline and week 72, 124 on tirzepatide and 36 on placebo
Tirzepatide lost 21.3% of bodyweight, placebo 5.3%. Of the weight lost, roughly 75% was fat and 25% was lean in both groups, and the proportion held across sex, age and the size of the weight loss. The ratio is not the drug's doing.
HumanBody composition changes during weight reduction with tirzepatide in the SURMOUNT-1 study of adults with obesity or overweight (2025)What this does not show
That the drug is therefore neutral for muscle. The same ratio applied to four times the weight loss is four times the lean tissue, and this trial measured lean mass by DXA rather than muscle directly.
Restricted to trials with big weight loss, the drug arms lost a bigger share as lean tissue
Measured in
1,334 participants across 21 randomised trials, restricted to interventions achieving at least 10% weight loss, measured by DXA, CT or MRI
The proportion of lost weight that was fat-free mass: 33.3% on incretin drugs, 34.2% after bariatric surgery, 22.3% on diet alone, and 7.7% on diet plus exercise. Exercise, not the drug and not the diet, is the variable that moves this number most.
ReviewEffects of incretin-based therapies, diet and exercise interventions, and bariatric surgery on fat-free mass in adults with overweight or obesity: a systematic review and meta-analysis (2026)What this does not show
Random assignment between these three routes. Nobody was randomised to surgery against a drug against a gym, so this pools arms from separate trials in different populations. It is also contested: a second 2026 meta-analysis, four times larger and without the 10% weight loss restriction, found no significant difference between incretins and lifestyle at all. Read the two together rather than picking one.
A larger meta-analysis found no difference between the drugs and lifestyle
Measured in
15,782 participants across 20 randomised trials of semaglutide, tirzepatide, liraglutide or intensive lifestyle intervention, body composition by DXA or MRI
Lean mass was 35.2% of weight lost on semaglutide, 25.4% on tirzepatide and 26.8% on liraglutide. Lifestyle intervention came out at 26.2%, and the difference against the drugs was not significant (p = 0.42). Adding resistance training to lifestyle produced the best figure in the analysis at 17.5%.
ReviewLean mass changes with incretin therapy versus lifestyle intervention: a systematic review and meta-analysis of randomised controlled trials (2026)What this does not show
Agreement with the other meta-analysis on this page, and that is the point of showing both. This one included trials at any amount of weight loss while the other required at least 10%, and they reach opposite conclusions about whether the drugs are worse than dieting. What they do agree on is the direction of the only lever either tested: resistance training produced the lowest share of weight lost as lean tissue in both. Note also that the 17.5% figure was measured without anybody taking a GLP-1, so applying it to a person on one is extrapolation.
When muscle was imaged instead of inferred, the loss tracked what bodyweight predicts
Measured in
246 adults with type 2 diabetes in the SURPASS-3 MRI substudy with a valid week 52 scan, thigh muscle volume quantified by MRI and contextualised against longitudinal UK Biobank data
Muscle volume fell, but broadly in line with the general relationship between bodyweight change and muscle volume rather than beyond it. Muscle fat infiltration, which is the quality measure, improved.
HumanTirzepatide and muscle composition changes in people with type 2 diabetes (SURPASS-3 MRI): a post-hoc analysis of a randomised, open-label, parallel-group, phase 3 trial (2025)What this does not show
A functional outcome. Volume and fat infiltration are structure, not strength, and this is a post-hoc exploratory analysis of an open-label trial funded by the manufacturer.
Adding protein to a very low calorie diet, without training, changed nothing
Measured in
108 adults with overweight or obesity randomised to 52 or 77 grams of protein a day on a 600 to 700 kcal diet for 8 weeks, lean mass by DXA
Lean mass fell about 6% in both arms and the share of weight loss that was lean tissue was identical at 27%. Resting metabolic rate fell 9 to 10% either way.
HumanA protein-supplemented very-low-calorie diet does not mitigate reductions in lean mass and resting metabolic rate in subjects with overweight or obesity: a randomized controlled trial (2021)What this does not show
That protein does not matter, and this is the most misread result in the area. Both arms were far below every intake on our protein page: 77 grams for a 100 kg adult is about 0.77 g/kg, which is the RDA. This trial compared a low intake with a slightly less low one and found no difference between them. It did not test a high protein intake, and nobody lifted anything.
Hair loss is real, and the base rate is 3.9%
Measured in
4,114 people taking GLP-1 receptor agonists across nine interventional studies, seven of them randomised
Pooled risk ratio 3.25 against placebo, 95% CI 1.44 to 7.36, and 3.59 when restricted to randomised trials in overweight or obesity. The single-arm event rate was 3.9%.
ReviewGlucagon-like peptide-1 receptor agonists and hair loss: a systematic review and meta-analysis (2026)What this does not show
A mechanism, or that the drug causes it directly rather than the speed of the weight loss doing so. A tripled risk on a 3.9% base is roughly three or four people in a hundred rather than a majority, and hair loss on this scale is well documented after rapid weight loss by any route, including surgery.
The thing nobody measured
Across thirty-five randomised trials of these drugs, not one reported an objective measure of physical function. Every figure argued over on this topic is a scan. Whether the lean tissue lost costs anybody strength, grip, walking speed or the ability to get out of a chair has, with a handful of incidental exceptions, simply not been tested.
Where function was measured for other reasons, it improved rather than declined, which is what you would expect when someone carrying a great deal of excess weight loses a lot of it. That is reassuring and it is not the same as having asked the question.
One trial is designed to answer the version of this that matters here. It randomises people on a GLP-1 to resistance training, to a higher protein target, to both, or to neither, with thigh muscle cross-sectional area by MRI as the primary outcome, and it is funded by nobody who sells any of this. It reports in 2029. Until then, anyone telling you confidently that protein fixes this, including us, is reasoning past the evidence rather than from it.
Questions people ask about this
- Do GLP-1 drugs cause muscle loss?
- They cause weight loss, and about a quarter of weight lost is lean mass whether the weight comes off on the drug or on placebo. The ratio is ordinary. What is not ordinary is the amount: in the SURMOUNT-1 scans the drug group lost 21.3% of bodyweight against 5.3% on placebo, so the same proportion means roughly four times the lean tissue.
- Is lean mass the same thing as muscle?
- No, and conflating them inflates every figure in this discussion. Lean body mass on a DXA scan includes organs, connective tissue, skin and body water as well as skeletal muscle, and muscle is roughly half of it. When thigh muscle was imaged directly by MRI rather than inferred from DXA, the volume lost tracked what the relationship between bodyweight and muscle predicts.
- Will eating more protein stop it?
- It is the reasonable thing to do and it has not been demonstrated for this situation. The trial usually cited against it compared 52 with 77 grams of protein a day, both around the RDA, and found no difference, which is a test of a low intake against a slightly less low one rather than a test of a high one. The trial that would actually answer this reports in 2029.
- What has the largest measured effect on lean mass during weight loss?
- Resistance training. Across trials reaching at least 10% weight loss, fat-free mass accounted for 22.3% of the weight lost on diet alone and 7.7% on diet plus exercise. No dietary variable measured so far moves the number that much.
- Is the hair loss permanent?
- The published randomised evidence establishes that the risk is raised, roughly tripled on a base rate of 3.9%, and does not establish a mechanism or a course. The pattern reported after rapid weight loss generally is a shedding phase that begins a couple of months in and recovers, and the condition that does scar, alopecia areata, has consistently not been found to increase. Anything more specific than that is not currently supported by the trial literature.