Semaglutide Loses a Third of Your Weight as Lean Mass. Tirzepatide Loses a Quarter. Your Arms Show It First.

The photograph is where most women notice

Not the scale, and not the mirror — the mirror you can angle. A photograph taken by somebody else, at a birthday, arm raised, and something about the upper arm reads differently than it used to.

The internet calls it Ozempic arms, which is the same vernacular that produced Ozempic face and Ozempic butt: a body part, a drug name, and no clinical definition behind either word.

There is real physiology underneath it, and it is not quite what the name implies. The arms are not being singled out by the drug. They are where a whole-body change becomes visible earliest, because the upper arm carries a thin layer over a muscle that most women were not deliberately loading, and because it is one of the few parts of yourself you routinely see photographed from an angle you did not choose.

What is actually being lost

This is where the research has moved recently, and the finding is more specific than the headlines about it.

A 2026 systematic review and meta-analysis pooled 20 randomized controlled trials covering 15,782 participants, restricted to trials that measured body composition by DXA scan or MRI rather than by estimate, with risk of bias assessed by Cochrane RoB 2 and certainty graded by GRADE¹.

It asked a precise question: of the total weight lost, what proportion was lean mass?

Two conclusions come out of that table, and the second one gets lost when this study is summarized.

The first is the reassuring one. Incretin drugs and lifestyle intervention lost a broadly comparable proportion as lean mass — the comparison returned p = 0.42, which is to say no detectable difference¹. The drugs are not doing something uniquely destructive to muscle. This is what losing a significant amount of weight does, by any method.

The second is the one worth knowing before you choose. Semaglutide's 35.2% and tirzepatide's 25.4% have confidence intervals that do not overlap. On this particular outcome the two drugs are not interchangeable, and the difference runs opposite to the way they are usually ranked — the drug that produces less total weight loss is losing a larger share of it as lean tissue.

That is a real consideration for a woman choosing between them, and it is almost never part of the conversation.

What only one thing changed

Look at the bottom row again, because it is the only intervention in the analysis that meaningfully moved the number.

Lifestyle plus resistance training brought lean-mass loss down to 17.5%, against 25–35% for everything else¹. The authors' own conclusion names the levers directly: muscle can be significantly preserved by integrating resistance training, adequate protein intake, and body-composition monitoring into weight-loss treatment¹.

Note what is not on that list. Not a particular drug, not a dose schedule, not a supplement. The thing that protected muscle was loading it.

For the arms specifically this is almost unfairly direct, because the upper arm is trivially easy to load and almost nobody does it. It does not require a gym membership or an hour. It requires resistance, twice a week, against something.

What the trial actually saw

The most detailed look at body composition on tirzepatide comes from a DXA substudy inside SURMOUNT-1 — 160 of the trial's 2,539 participants, scanned at baseline and again at week 72. Notably for this readership, 73% were women².

Roughly 75% of the weight lost was fat and 25% was lean — for the drug group and the placebo group alike, and those proportions held across subgroups by sex, age and how much weight was lost².

Two caveats belong with those numbers. The substudy is 160 people out of 2,539, which is a small window onto a large trial. And several authors are employed by Eli Lilly, which makes tirzepatide — worth knowing, though a finding that a quarter of the loss is lean tissue is not a flattering one to manufacture.

The ratio being identical in the placebo arm is the most useful thing in the table. It says the composition of the loss is a property of losing weight, not a property of the drug.

Why the arm and not somewhere else

Here it is worth being honest about the limits of the evidence, because no study has measured arms separately. Neither of these papers reports regional DXA results.

What can be said is anatomical rather than experimental. The upper arm has a relatively thin subcutaneous layer over the triceps, so a change underneath shows through sooner than it would somewhere padded. It is rarely loaded in ordinary daily life in the way legs are by walking and stairs. And it is uniquely exposed to your own gaze in photographs, from angles nobody chooses for themselves.

That combination — visible early, unloaded by default, photographed constantly — is enough to explain why the arms are the part that gets named, without needing the drug to be doing anything regional at all.

One thing it is not: the injection site. Both labels list the upper arm among the acceptable places to inject, and Wegovy's pharmacokinetics section reports similar drug exposure whether it is given in the abdomen, thigh or upper arm. Injecting into your arm is not what changed your arm.

The part that is not physiology

A qualitative study in JAMA Network Open interviewed 30 people across 15 US states, 19 of them women, mean age 54, and analyzed what they said without a predetermined framework³.

Two of its findings sit directly on this subject. Participants were clear that these drugs are not a standalone weight-loss solution. And they described stigma shaped by why they had been prescribed it³.

That second one is the quiet weight this subject carries. An arm that has changed shape is legible to other people in a way a number on a scale is not. It is the part of the transformation that gets commented on — approvingly, usually, which does not necessarily make it easier — and the part that announces to a room that something is going on before you have decided who to tell.

Nobody prepares women for the fact that visible change is a form of disclosure.

What this means practically

Ask which drug, and ask about muscle, before you start. The 35.2% against 25.4% difference is real and is not part of most prescribing conversations.

Start resistance training before you need it. It is the only variable in a 15,782-person analysis that changed the outcome, and it works better as prevention than as repair.

Do not read the arm as evidence of a mistake. A quarter of the loss being lean tissue happened in the placebo group too. It is the cost of the loss, not a sign the drug is harming you.

And if what actually bothers you is the photograph rather than the physiology — that is worth saying out loud rather than routing through a question about muscle. The two problems have different solutions, and only one of them is solved by lifting something heavy.