am i losing muscle on this.

Published September 7, 2026·Training · Body composition·16 min read

GLP-1 muscle loss, measured: on semaglutide and tirzepatide, between a quarter and a third of the weight that comes off is lean mass. Lifting three days a week plus a protein floor is what keeps it, in trials. Here is the lifting half, with the numbers.

you started the shot, the scale finally moved, and somewhere around week ten you noticed the thing nobody on the pharmacy leaflet mentioned. the jeans fit. the arms look smaller. the pull-up that was almost there is now further away than it was at the heavier weight. and then you read a headline that said up to 40 percent of what you lose on these drugs is muscle and you did the maths on your own kilos and felt sick.

i do not sell a GLP-1 and i am not going to tell you whether to take one. that is between you and whoever writes the script. what i can do is the half of this that every clinic page skips: the numbers from the scans, why the muscle goes, and the exact lifting and protein protocol that kept it in the trials. the short version is at the top. the long version is why you should believe it.

how much of what i lose is actually muscle.

start with the two trials that put people in a DXA scanner, because everything else is a review of them. DXA is the low-dose x-ray that splits you into fat mass, lean mass and bone. in STEP 1, the trial that got semaglutide 2.4 mg approved for weight, 140 participants were scanned at the start and at 68 weeks. body weight fell 15.0 percent. total fat mass fell 19.3 percent. total lean mass fell 9.7 percent. run those against the starting body composition the paper reports [43.4 percent fat, 53.9 percent lean] and the lean loss works out to about 5.2 percent of starting weight against 8.4 percent from fat. that is 35 to 38 percent of the loss coming from lean tissue, depending on how you count the bone. a third, honestly rounded. that is my arithmetic on their published percentages, not a number they print.

in SURMOUNT-1, the tirzepatide trial, 160 people were scanned over 72 weeks. weight fell 21.3 percent, fat 33.9 percent, lean 10.9 percent, and the authors did the split themselves: about 75 percent of the weight lost was fat and 25 percent was lean. the placebo group lost far less but split the same way. so did every subgroup they checked, including women, who were 73 percent of the scanned sample.

now the honesty line, because the headlines skip it. lean mass is not muscle. it is everything that is not fat or bone: skeletal muscle, yes, but also organ tissue, blood volume, glycogen and the water bound to it, and the water inside fat cells, which gets counted as lean and leaves when the fat does. a 2024 review in Diabetes, Obesity and Metabolism found lean loss reported anywhere from 40 to 60 percent of weight lost in some trials down to 15 percent or less in others, and pointed out that the MRI studies, which measure actual muscle volume, find reductions in line with what the weight loss alone would predict. the headline "40 percent is muscle" is a lean-mass number wearing a muscle costume.

put the table together and the picture is boring in the way real answers usually are. lose a lot of weight fast, by any method, and somewhere between a fifth and a third of it is lean. a 2019 review of every GLP-1 body-composition study put it at 20 to 50 percent in over half the trials and called it "consistent with diet-induced weight loss and bariatric surgery". the drug is not special. the speed is.

why does the drug take muscle at all.

it does not, directly. three things it causes do. first, the deficit. in the study that measured what people actually ate on semaglutide, 30 adults ate 24 percent less at free-choice meals, about 725 calories a day, with no change in resting metabolic rate once you adjust for lean mass. that is a large deficit that arrives without you planning it. muscle is metabolically expensive tissue, and a body in a deep deficit with no reason to keep it will spend it.

second, the protein goes with the appetite. this is the part i think matters most and nobody on page one has read. in the CRAVE study, published this June, people on semaglutide or tirzepatide averaged 0.90 grams of protein per kilogram of body weight at 24 weeks, and higher absolute protein intake tracked with better preservation of estimated muscle mass [correlation 0.41, borderline significant, 23 people, so hold it loosely]. in a 2026 cohort of 387 adults from online GLP-1 support groups who kept food diaries, average intake was 753 calories and 33 grams of protein a day, and fewer than 10 percent met the protein recommendation. that is self-reported and from support groups, so treat it as the worst case rather than the average. but 33 grams is one chicken breast. you cannot keep muscle on one chicken breast.

third, you move less, and you do not notice. at the Endocrine Society meeting in June, a team pulled wearable data from 753 adults in the NIH All of Us program who started a GLP-1. daily steps fell from 5,047 to 4,487. moderate-to-vigorous activity fell from 28 to 22 minutes a day. the cohort was 78.6 percent female, average age 52.7. the lead author's line was that "exercise cannot be optional for people taking these medications". it is a conference abstract, not a peer-reviewed paper yet, and it cannot say why. but the direction is the opposite of what everyone assumes, which is that lighter people move more.

stack those three. a 700-calorie deficit, protein at about half of what a deficit needs, and less movement than before. there is no mechanical signal telling the body the muscle is in use, and there are not enough amino acids to rebuild what training would break down anyway. muscle is lost by absence, not by attack. which is good news, because absence is the easy thing to fix.

one open question, flagged honestly: an April 2026 preprint of nearly 8,000 clinic patients, mostly measured by bioimpedance rather than DXA, found tirzepatide lost more lean mass than semaglutide at every point, 5.6 versus 3.6 percent at 12 months, alongside more total weight. it is unreviewed and observational and the measurement method is noisy. i mention it so you know it exists, not so you switch drugs over it.

does it actually matter, if my doctor says my function is fine.

the medical argument here is real and i am not going to pretend it is one-sided. in JAMA in 2024, three of the most careful people in this field wrote that "the recent concern that marked weight loss induced by GLP-1-based anti-obesity medications can cause physical frailty or sarcopenia is not supported by data". their point: the lean loss relative to your starting body is small, it is smaller than the fat loss, and physical function usually improves because you are carrying less. the SEMALEAN study backs that up. 106 people with severe obesity, a year on semaglutide, lean mass down 3 kilograms by month seven then flat, and grip strength up 4.1 kilograms. sarcopenic obesity, the state of too little muscle for your size, fell from 49 to 33 percent of the group. no control group, one hospital, but the direction is clear.

the other side, in Diabetes Care the same year: incretin drugs cause "rapid and significant loss of lean mass, about 10 percent or about 6 kilograms, comparable to a decade or more of aging", and supervised lifting for more than ten weeks adds about 3 kilograms of lean mass and 25 percent strength in both men and women. same data. different question.

and that is the thing. the medical question is "will this make you frail." your question is different. you are reading a training site at one in the morning because you did not start the shot to become a smaller, softer version of the same shape. you wanted the shoulders to show up when the fat left. you wanted the pull-up. the JAMA authors are right that you will probably not be frail. they are not claiming you will be built, and nothing in the scan data says you will be.

there is a second reason it matters, and it is about what happens after. in the STEP 1 extension, 327 people were followed for a year after semaglutide was withdrawn. they regained two-thirds of what they had lost. and the regain is not symmetrical. in 78 postmenopausal women tracked through a diet and the year after it, 0.26 kilograms of lean tissue was lost for every kilogram of fat lost, but only 0.12 kilograms of lean came back for every kilogram of fat regained. that is a ratchet. each cycle of lose-and-regain leaves you with a bit less muscle and a bit more fat at the same weight. if the shot is a season rather than a life, the muscle you keep during it is the only thing that is still yours afterwards.

what does lifting actually do to the number.

this is the part that should be on every clinic page and is not, because the trials were done on dieters rather than on people taking the drug. that matters less than it sounds. the drug produces a deficit. these trials study what lifting does inside a deficit.

the cleanest one is Villareal 2017 in the New England Journal. 160 obese older adults, six months, all on the same diet losing 9 percent of body weight, randomised to aerobic exercise, resistance training, both, or nothing. lean mass fell 5 percent in the aerobic group, 3 percent in the combined group and 2 percent in the resistance group. strength rose 4 percent with cardio, 18 percent combined, 19 percent with lifting. hip bone density fell 3 percent with cardio and half a percent with lifting. same weight lost. entirely different body at the end.

it is not one trial. a 2018 meta-analysis pooled six randomised trials of obese older adults dieting with or without lifting three times a week for 12 to 24 weeks. resistance training removed 93.5 percent of the lean mass the diet would otherwise have cost, about 0.8 kilograms, while fat loss and total weight loss were no different from dieting alone. the ratio of strength to lean mass, a rough measure of muscle quality, went up 20.9 percent with lifting and down 7.5 percent without it. the authors' conclusion: lifting during a diet "is able to prevent almost 100 percent" of the muscle loss. older adults, again, so the fine print is that a younger body has more to work with, not less.

and at the far end of what is possible: Longland 2016 put 40 young men on a 40 percent calorie deficit, which is roughly the deficit a GLP-1 hands you, with six days a week of lifting and intervals. the half eating 2.4 grams of protein per kilogram gained 1.2 kilograms of lean mass while losing 4.8 kilograms of fat in four weeks. the half at 1.2 grams per kilogram held lean mass flat and lost 3.5 kilograms of fat. men, young, supervised, brutal, four weeks. it does not transfer to you one-to-one. it does prove the ceiling: in a deficit that size, with enough protein and enough tension, the lean line can go up.

what about the drug and the lifting together.

one trial has done exactly that, with liraglutide, the older daily cousin of semaglutide. S-LiTE, New England Journal 2021: 195 adults lost 13.1 kilograms on an eight-week diet, then spent a year on placebo, supervised exercise, liraglutide, or both. against placebo, exercise held an extra 4.1 kilograms off, liraglutide 6.8, and the combination 9.5. the combination cut body-fat percentage by 3.9 points, about twice the exercise group, and it was the only arm that improved blood sugar, insulin sensitivity and fitness at the same time. the drug alone did not do that. the exercise alone did not do that.

then they turned everything off. one year after treatment ended, the people who had been on liraglutide alone regained. the people who had exercised, with or without the drug, mostly did not. weight regain was 6.0 kilograms larger after stopping the drug than after stopping the exercise program, and the paper's own summary line is that body weight and composition "were maintained one year after termination of supervised exercise, in contrast to weight regain after termination of treatment with obesity pharmacotherapy alone". the exercise did something the drug could not, which is change what your body does when nobody is intervening.

the industry knows this is the number. in March, Nature Medicine published BELIEVE, a 507-person trial pairing semaglutide with bimagrumab, an antibody that blocks the receptor that tells muscle to stop growing. weight loss at 48 weeks was 14.2 kilograms on semaglutide alone and 17.8 on the pair. the lean numbers, as reported by the lead centre and at the diabetes association meeting: semaglutide alone lost 7.4 percent of lean mass and 71.8 percent of its loss was fat; the combination lost 2.9 percent lean and 92.8 percent of its loss was fat. bimagrumab is investigational, an intravenous infusion, and not something you can get. i bring it up because a pharmaceutical company just ran a phase 2 trial to buy back the muscle that lifting keeps for free, and got 92.8 percent. the lifting version in the pooled diet trials got 93.5.

the small studies point the same way. a 2025 case series followed three people on semaglutide or tirzepatide who lifted three to five days a week and ate 1.6 to 2.3 grams of protein per kilogram of lean mass. they lost 33, 27 and 13 percent of body weight. two of them gained lean tissue while doing it. three people is not evidence of anything except that it can be done. the randomised trials are what say it usually is.

how much protein, when i'm not hungry.

the floor is the number from the largest meta-analysis of protein and lifting, 49 trials and 1,863 people: gains in lean mass stop increasing at about 1.6 grams per kilogram of body weight per day. that is for people eating enough. in a deficit, the evidence-based guidelines for people trying to keep muscle while getting lean, written for natural bodybuilders, use 2.3 to 3.1 grams per kilogram of lean mass and a loss rate of 0.5 to 1 percent of body weight a week. if you have more to lose, the per-lean-kilo version is kinder than the per-total-kilo one, because it does not ask you to feed the fat.

worked example, because a range is not a plan. you weigh 90 kilograms. 1.6 times 90 is 144 grams a day. if a scan or a decent estimate puts your lean mass at 55 kilograms, 2.3 times 55 is 127 grams. call it 120 to 145. now look at what the drug cohorts actually eat: 0.90 grams per kilogram is 81 grams for that person, and the support-group diaries averaged 33. the gap between what keeps muscle and what people on the shot are eating is 40 to 110 grams a day. that is the whole mechanism of this article in one line, and it is why the clinic advice "eat more protein" fails: it is a direction, not a dose.

the how is not glamorous. protein first on every plate, before the thing that fills you. 30 to 40 grams a feeding, three or four feedings, which is the shape that lets a muscle protein synthesis response happen each time rather than one heroic dinner your gut refuses. on a drug that slows the stomach, a liquid feeding counts: a shake over ice at two in the afternoon is not cheating, it is lunch. and a warning the guidelines themselves carry: if you are losing faster than about 1 percent of body weight a week, the muscle-keeping maths gets hard no matter what you eat. that is a conversation with your prescriber about pace and dose, not with me.

the protocol.

this is what i would hand you if you were on the shot and DMed me. it is not fancy. the trials that kept muscle used three full-body sessions a week, compound lifts, and enough weight that the last rep was hard. that is all this is.

so. do i have to lift.

i spent years at 1,200 calories a day on purpose, long before any of this had a brand name. i know exactly what a body does when the food goes away and the lifting does not come with it. it gets smaller everywhere and softer in the places you wanted hard, and the scale congratulates you the whole way down. nobody told me the number was the wrong number. i am telling you.

the drug is doing one job, and by the trial data it does it well. it is not doing the other job, and by the same trial data, nothing about it will. three sessions, a protein number instead of a protein vibe, the steps you had before, and a lift you test once a month. that is the version of this where the shoulders show up when the fat leaves.

and if you are reading this at one in the morning because the arms looked smaller in the mirror tonight: they probably are, a little. that is not the end of anything. that is the week you start.

Quick answers

does ozempic or zepbound cause muscle loss?

the weight loss does, and the drug causes the weight loss. in the semaglutide trial that scanned people, lean mass fell 9.7 percent over 68 weeks and made up roughly a third of the weight lost. in the tirzepatide trial it was about a quarter. lean mass includes water, organ tissue and fluid, not only muscle, and reviews find the muscle portion is in line with any weight loss of that size, not a drug-specific effect. the drivers are the calorie deficit, protein intake that falls with appetite, and activity that drops after starting. all three are fixable.

how much protein should i eat on a glp-1?

a floor of about 1.6 grams per kilogram of body weight per day, which is where the largest meta-analysis of protein and lifting found gains stop increasing. in a deficit with training, 2.3 to 3.1 grams per kilogram of lean mass is the range used in evidence-based cutting guidelines. for a 90 kilogram person that is 120 to 145 grams a day, eaten protein-first across three or four feedings of 30 to 40 grams. people on these drugs average about 0.9 grams per kilogram, and in one cohort fewer than 10 percent met the recommendation.

will lifting weights slow down my weight loss on a glp-1?

no, it changes what the weight is made of. in a trial of 160 older adults dieting to a 9 percent loss, the resistance-training group lost the same weight as the aerobic group but kept 3 percentage points more lean mass and gained 19 percent strength. in the liraglutide plus exercise trial, the combination lost more weight than the drug alone and twice the body fat of exercise alone. in pooled trials, lifting three times a week removed 93.5 percent of the lean loss a diet caused without changing fat loss.

what happens to my muscle if i stop the shot?

the weight tends to come back and the muscle mostly does not. one year after stopping semaglutide, trial participants regained two-thirds of what they lost. in postmenopausal women followed through regain, 0.26 kilograms of lean tissue was lost for every kilogram of fat lost but only 0.12 kilograms of lean regained per kilogram of fat regained. the exception is exercise: a year after stopping both liraglutide and a supervised exercise program, the people who had exercised kept their weight and body composition while the drug-only group regained.

keep reading

• how much protein do i actually need. — the per-kilo number, the per-feeding number, and the studies behind both
• when am i supposed to add weight. — the smallest-plate rule from the protocol table, in full
• why did i gain weight overnight. — the water and glycogen behind the scale, which matters twice as much on a drug that moves it fast
• how long does it take to build visible muscle. — the timeline for the shoulders once the fat is out of the way
• is creatine just a bro thing. — the other 5 grams a day, and the meta-analysis from this month

---

References

educational content, not medical advice. the numbers above are group results: the DXA substudies are 140 and 160 people; the lifting-during-a-diet trials are mostly older adults and the 2.4 g/kg trial is young men; the protein-intake cohorts are small or self-reported; the BELIEVE lean-mass shares come from conference reporting rather than the paper's abstract; the activity data is a conference abstract; the tirzepatide-versus-semaglutide comparison is an unreviewed preprint. starting, stopping, dosing or pacing a GLP-1 is a decision for you and your prescriber. everything in the tables is sourced.

- MoMo