Does Ozempic cause muscle loss? What the STEP 1 scans showed

Does Ozempic cause muscle loss?

Semaglutide users lose lean mass as well as fat, and the STEP 1 body scans put lean mass at about 40% of the tissue lost.

That figure is a scan measure of lean mass, not a biopsy of muscle. The trial gave semaglutide 2.4 mg a week for 68 weeks, the dose sold as Wegovy. Ozempic is the same drug at lower doses, and the only lower-dose scan data found for this page (SUSTAIN 8, 1.0 mg) are small.

The drug does not appear to attack muscle directly. Large weight loss from any cause lowers lean mass, and the questions are why, how much, and whether the share is unusual. The sections below give the numbers, the arithmetic and the mechanism.

What did the STEP 1 scans show?

In the STEP 1 DXA substudy, 140 adults on semaglutide or placebo were scanned at the start and at week 68, and lean mass fell 9.7% on semaglutide.

DXA is dual-energy X-ray absorptiometry, a low-dose scan that splits the body into fat, bone and everything else, called lean mass. The table collects what was reported.

Item	Semaglutide 2.4 mg	Placebo
Participants in DXA substudy	95	45
Whole STEP 1 trial	1,961 adults, 68 weeks, plus lifestyle advice	same
Body weight change	-15.0%	-3.6%
Total fat mass change	-19.3%	no major change
Lean mass change	-9.7%	no major change
Lean mass as share of body mass	up 3.0 percentage points	no major change

Source: Journal of the Endocrine Society, 2021 (exploratory analysis, abstract). Participants had a BMI of 40 or below and came from 9 sites. Average starting weight was 98.4 kg (217 lb), BMI 34.8, 76% women. Only some of the 1,961 trial participants had scans, so the substudy is about 7% of the trial.

The same data are reported in kilograms in a 2026 review of the NEJM paper. In the trial-product estimand, which counts people while they stayed on treatment, fat fell 10.40 kg (22.9 lb) and lean mass fell 6.92 kg (15.3 lb).

How much of the weight lost on Ozempic is muscle?

About 40% of the DXA-measured loss in STEP 1 was lean mass, which is an upper bound for muscle because lean mass includes water and organs.

Here is the calculation using the kilogram figures:

1. Fat lost: 10.40 kg.
2. Lean lost: 6.92 kg.
3. Total DXA tissue lost: 10.40 + 6.92 = 17.32 kg.
4. Lean share: 6.92 ÷ 17.32 = 0.40, or 40%.

The percentage figures give a similar answer. This is my own arithmetic from the published starting proportions, so treat it as a check, not a result:

1. Starting fat: 43.4% of 98.4 kg = 42.7 kg. Falling 19.3% means about 8.2 kg lost.
2. Starting lean: 53.9% of 98.4 kg = 53.0 kg. Falling 9.7% means about 5.1 kg lost.
3. Lean share: 5.1 ÷ (8.2 + 5.1) = 0.38, or 38%.

The two versions differ in absolute kilograms because they use different analysis methods. Both land near 40%. The headline claim that “40% of Ozempic weight loss is muscle” is therefore a fair reading of the lean share, and an overreading of the muscle share.

Where does the “up to 40%” figure come from?

The viral “up to 40%” claim matches the STEP 1 lean share, but no post making it shows a table of its own, so I cannot say which source it used.

The Mayo Clinic Store education page says “approximately 25% to 40%” of weight loss during GLP-1 therapy may come from lean mass, with no citation next to the range. The 2026 Diabetes & Metabolism Journal review words it as “often ~25-40%”. The same review lists a second semaglutide trial, SUSTAIN 8 (1.0 mg, 52 weeks, type 2 diabetes), at fat -3.4 kg and lean -2.3 kg, which is 2.3 ÷ 5.7 = 40.4%.

How much lean mass is lost across GLP-1 trials?

Across body-scan analyses the lean share of weight lost runs from about 14% to 60%, and most large obesity-dose results sit between 25% and 40%.

The table recalculates each share from the kilograms in the 2026 Diabetes & Metabolism Journal review, except SURMOUNT-1, which is from the trial paper. Share = lean ÷ (fat + lean). These are secondary figures, not re-read from each trial.

Analysis	Drug and length	Fat lost (kg)	Lean lost (kg)	Lean share
STEP 1 substudy	Semaglutide 2.4 mg, 68 wk	10.40	6.92	40%
SUSTAIN 8 substudy	Semaglutide 1.0 mg, 52 wk, diabetes	3.4	2.3	40%
SURMOUNT-1 substudy	Tirzepatide 5-15 mg, 72 wk	15.9	5.6	26%
SURMOUNT-1 placebo arm	Placebo plus lifestyle, 72 wk	3.6	1.2	25%
BELIEVE control arm	Semaglutide 2.4 mg, 48 wk	9.6	3.9	29%
EMBRAZE control arm	Tirzepatide, 24 wk	8.0	3.5	30%
LEAD-2 substudy	Liraglutide 1.8 mg, 26 wk, diabetes	2.4	1.5	38%
LEAD-3 substudy	Liraglutide 1.8 mg, 52 wk, diabetes	1.0	1.5	60%
Astrup et al.	Liraglutide 3.0 mg, 20 wk	6.8	1.1	14%

Two things stand out. The 40% figure is a real result, but it sits near the top of the range for obesity doses. And the highest share (LEAD-3, 60%) comes from a trial where fat fell only 1.0 kg, so a small fat loss inflates the percentage. The review does not give sample sizes for most of these rows.

What does that mean in kilograms for a 100 kg person?

For a 100 kg person who loses 15% of body weight, a 25% lean share is 3.75 kg, and a 40% share is 6.0 kg of lean mass.

This is arithmetic on an example, not a prediction for you. The 15 kg is chosen because STEP 1 participants lost 15.0% of body weight on semaglutide.

Lean share	Lean mass lost from 15 kg	If muscle is 50% of it	If muscle is 40% of it
25% (SURMOUNT-1 range)	3.75 kg (8.3 lb)	1.9 kg	1.5 kg
30% (BELIEVE, EMBRAZE)	4.5 kg (9.9 lb)	2.3 kg	1.8 kg
40% (STEP 1)	6.0 kg (13.2 lb)	3.0 kg	2.4 kg

The last two columns use the reference ratios in the next section (skeletal muscle about 50% of fat-free mass in the reference man, 40% in the reference woman). Those ratios describe total stores, not what is lost, so the muscle columns are an illustration. No GLP-1 trial has measured the true split.

The same lean loss also looks different against the starting stock. In SURMOUNT-1 the lean loss of 5.6 kg was 10.9% of 50.8 kg starting lean mass, while fat mass fell 33.9%.

Is lean mass the same as muscle?

No. DXA lean mass is all non-fat, non-bone tissue, so it counts water, organs and skin as well as skeletal muscle.

A 2024 review by Tinsley and Heymsfield gives the makeup of fat-free mass as about 74% water, 19% protein and 6.5% mineral. Skeletal muscle is about half of fat-free mass in the reference man (28 kg of 56.5 kg) and about 40% in the reference woman (17 kg of 42 kg).

The same review describes one weight-loss study in women where 1.5 kg of fat-free mass was lost out of 9.5 kg total. Skeletal muscle was 0.9 kg of that 1.5 kg, but most of the muscle loss was water inside the muscle (0.7 kg), and only about 0.15 kg was muscle protein. Kidney, heart and liver accounted for 0.1 kg. That was one low-energy diet study, not on a GLP-1 medicine, and it shows why a lean-mass number can overstate muscle tissue lost.

Why does Ozempic cause muscle loss?

Ozempic lowers lean mass mainly because it causes a large energy deficit, and the body covers that deficit from fat and some lean tissue unless protein and training give muscle a reason to stay.

1. The drug lowers appetite. GLP-1 drugs act on brain regions that control hunger and fullness. The 2025 joint advisory from four US nutrition and obesity societies reports energy intake 16% to 39% lower than placebo.
2. A deficit opens. Most stored energy drawn on comes from fat. Some comes from lean mass, and the advisory says the amount depends on how deep the calorie restriction is, how fast weight falls, and whether people strength train.
3. Protein often falls with it. A small appetite usually means less protein as well as fewer calories. The advisory says low protein from reduced appetite may add to muscle loss, particularly in older adults, in menopause, in sedentary people and in people who do not strength train.
4. Muscle gets less of a reason to stay. Without resistance training, the signal to hold on to muscle is weaker. Whether GLP-1 drugs change how much people move is not shown by the sources used here.

A 2026 review of the molecular studies describes a paradox. Lab data show protective signals in muscle, yet trials show lean mass falling. The authors suggest a systemic energy and protein deficit as the explanation rather than a direct drug effect, and add that current evidence does not fully confirm it.

The clearest test is a 2026 mouse experiment in Cell Reports Medicine. The researchers immobilized one hind leg in obese mice for 7 days (8 mice per group). Mice on semaglutide, mice on a GLP-1/glucagon drug and mice fed only the same amount as the semaglutide group lost similar muscle. Muscle loss depended on how much weight came off, not on whether the drug or reduced food caused it. One caution: the semaglutide and pair-fed mice had different muscle protein profiles in a smaller analysis (4 mice per group), so the drug is not identical to dieting at the molecular level, and the meaning of that difference is unconfirmed.

What do mouse and human studies each show?

Mouse studies show fat falling faster than muscle, and human studies show the same pattern on scans, though neither type of study proves harm or safety in a given person.

Study	Who	Size and length	Main finding	What it cannot show
Cell Rep Med 2026, tirzepatide	Obese male mice	5 per group, 14 days	Weight -35%, body fat -73%, lean body mass -13%. Lean was about 20% of weight lost	Anything about people, or about women
Cell Rep Med 2026, semaglutide	Obese male mice	10 per group, 28 days	Weight 22% below obese controls. Strength relative to body size held	Long-term effects
Cell Rep Med 2026, immobilization	Obese male mice	8 per group, 7 days	Muscle loss matched weight loss in drug and pair-fed groups	Real-life dieting, which differs from pair-feeding
Cell Rep Med 2026, clinical trial	Adults with obesity and type 2 diabetes	About 10 per group, 12 weeks of semaglutide	About 70% fat and 30% lean lost. Grip and knee-extension strength unchanged	Firm conclusions. The authors call it small and underpowered
STEP 1 DXA substudy, 2021	Adults without diabetes	140 (95 semaglutide 2.4 mg), 68 weeks	Lean mass -9.7%, lean share of body mass +3.0 points	Strength or function
Meta-analysis, Int J Obes 2026	Adults in obesity-dose RCTs	7 studies, 821 people	Lean mass -1.74 kg (95% CI -3.04 to -0.45), but +1.81 points as a share of body weight. I² 98%	A single answer, since studies disagree widely

The 2026 International Journal of Obesity meta-analysis pooled 5 liraglutide trials, 1 semaglutide trial and 1 tirzepatide trial that could not be pooled. The single semaglutide trial gave -5.44 kg, or -9.9%. No included study measured strength or physical capacity. A second 2026 meta-analysis in Diabetes, Obesity and Metabolism covered 19 studies, rated certainty low for adiposity and very low for muscle quantity, and called its findings hypothesis-generating.

Is the muscle loss on Ozempic worse than losing weight by dieting?

Diet-only weight loss also costs lean mass, and the commonly cited rule is that about one-fourth of the weight lost is fat-free mass.

The table puts the semaglutide result next to other numbers. These come from different populations, scanning methods and trial lengths, so the comparison is rough.

Setting	Lean (or fat-free) share of weight lost	Source
Semaglutide 2.4 mg, STEP 1 DXA substudy, 68 weeks	about 38-40%	J Endocr Soc 2021; Diabetes Metab J 2026
Tirzepatide, SURMOUNT-1 DXA substudy, 72 weeks	about 25%	Diabetes Obes Metab 2025
Placebo arm (lifestyle advice), SURMOUNT-1	about 25%	Diabetes Obes Metab 2025
“Quarter rule” for dieting	about 25% (origin: 0.22 in women aged 14-60)	Obesity Reviews 2014
Low-calorie diets, median (Chaston review)	14%	J Endocr Soc 2024
Very low-calorie diets, median (Chaston review)	about 23%	J Endocr Soc 2024
Sedentary diet-only, pooled 16-week trials	27%	Obesity Reviews 2014

Three points follow. The STEP 1 share sits above the usual diet-only range. The SURMOUNT-1 share does not, and its placebo arm shows 25% lean too. And the Obesity Reviews authors say the fraction is not constant. It is higher in the first days or weeks and when baseline fat is lower, and the review calls the quarter rule “at best an approximation”.

The Diabetes & Metabolism Journal review also cites a meta-analysis of incretin drugs against lifestyle: lean share 35.2% for semaglutide, 25.4% for tirzepatide, 26.8% for liraglutide, 26.2% for lifestyle alone and 17.5% for lifestyle plus resistance training. The review gives no year or sample size for it, and I could not open the original. Why STEP 1 and SURMOUNT-1 differ is not answered by these data. Different people, scanners, drug and weight-loss size are all candidates.

Does Wegovy or Mounjaro change the answer?

Wegovy is the same 2.4 mg dose tested in STEP 1, so the 38-40% figure applies to it. Tirzepatide (Mounjaro, Zepbound) data are different.

In the SURMOUNT-1 DXA substudy (n = 160: 124 on tirzepatide, 36 on placebo, 72 weeks), body weight fell 21.3%, fat mass 33.9% and lean mass 10.9% on tirzepatide. Placebo users lost 5.3% of weight and 2.6% of lean mass. About 75% of weight lost was fat and 25% lean, in both arms. A worked check with the average starting weight of 102.5 kg: 21.3% is about 21.8 kg lost, of which a quarter is about 5.5 kg lean and three quarters about 16.4 kg fat.

The placebo group lost about the same share as lean, with far less total weight. That fits the deficit explanation: the share is similar, and a bigger loss means more kilograms of lean mass. The full tirzepatide data are on the page about Mounjaro and Zepbound scans.

Can you keep muscle while taking Ozempic?

In diet trials, adding exercise cut the share of weight lost as fat-free mass, though STEP 1 did not test this.

In pooled 16-week diet trials, the fat-free share was 0.27 for sedentary diet-only, 0.17 with resistance training and 0.13 with aerobic exercise. Those are diet studies, not GLP-1 studies. The 2025 advisory points to enough protein and resistance training together, says protein alone is unlikely to prevent loss, and discusses strength training at least 3 times a week plus 150 minutes of moderate aerobic activity a week.

Protein targets are covered in how much protein to eat on a GLP-1, a step-by-step plan is in how to prevent muscle loss on Ozempic, and what training can and cannot do is in can you build muscle on Ozempic. Nothing here is a plan for you. Your prescriber can tell you what suits your health. For the full set of guides, see the guides index.

Is GLP-1 muscle loss a problem?

Lean mass loss on GLP-1 drugs is a concern for some people, but the trials with strength tests found little change in strength, so a scan number alone does not show harm.

Sarcopenia means low strength and function, not just low mass, and the larger scan trials did not measure that. People who are older, already weak, or who eat very little have the most reason to take it seriously. The advisory names older age, menopause, lower testosterone and sedentary behavior as risk factors.

What the scans cannot tell you

The scan data come from small substudies, mostly funded by the drug makers, and they measured how much mass was lost, not whether anyone became weaker.

• STEP 1 scanned about 140 people, 76% women, at 9 sites; SURMOUNT-1 analysed 160. Results may not apply to you.
• STEP 1 was an exploratory analysis with several authors employed by Novo Nordisk. SURMOUNT-1’s substudy was funded by Eli Lilly.
• The percentage and kilogram figures come from two different reports and two analysis methods. The kilogram figures come from a 2026 review, not from the NEJM paper itself, which I could not open on 2026-10-26.
• DXA cannot separate muscle from water or organs.
• Whether the drugs have any direct effect on human muscle is open. The mouse protein difference is unexplained.
• No scans followed people after treatment stopped, so whether the loss reverses is not covered.

Where the data are thin, the honest answer is that some lean mass is lost, it is probably less than 40% muscle, and function matters more than a scan number. See your clinician if you notice weakness, falls or trouble climbing stairs, and do not change or stop a medicine on your own.