Weight loss of any kind takes some lean tissue with the fat. The question that matters is whether these drugs take more than a diet producing the same loss would, and the published record is thinner and less certain than the confidence of the argument around it suggests.
The best-controlled figures available
A prespecified analysis of the SMART trial measured body composition by bioimpedance spectroscopy in 101 adults with chronic kidney disease, overweight or obesity and no type 2 diabetes, randomized to semaglutide 2.4 mg weekly or placebo for 24 weeks. Against placebo, semaglutide changed total body weight by −9.1 kg (95% CI −11.0 to −7.2), lean body mass by −2.5 kg (95% CI −6.6 to 1.6) and fat mass by −3.9 kg (95% CI −7.8 to 0.0)[1].
Read the intervals rather than the point estimates. The lean mass interval runs from −6.6 to +1.6 kg, which includes no change at all, and the fat mass interval reaches zero at its upper end. The study is small, it is short, and its population has kidney disease. It is the most carefully controlled measurement available and it does not settle the question. A reader who has seen a confident figure for how much muscle these drugs cost has seen something the data does not support.
What the large trials measured instead
STEP 1 randomized 1,961 adults and reported a mean weight change of −14.9% against −2.4% on placebo over 68 weeks [2]. Body composition was not a coprimary endpoint, and the trial’s reported outcomes are weight, weight-loss thresholds and cardiometabolic risk factors. The same holds for the withdrawal extension, which followed weight and cardiometabolic variables after the drug stopped and did not report what happened to lean tissue.
An observational analysis of tirzepatide published in 2026 reports body weight and body composition changes in adults with overweight and obesity [3]. It is not randomized, so it describes what happened to the people it followed rather than what the drug caused.
Why the question is hard to answer at all
Lean body mass is not muscle. Bioimpedance and dual-energy absorptiometry both measure fat-free tissue, which includes water, glycogen and organ mass. Semaglutide reduced extracellular water by 0.9 L (95% CI −1.6 to −0.1) in the same SMART analysis[1]. Some of what a scan records as lost lean mass is fluid, and no bedside method separates the two.
The second difficulty is the comparator. Any intervention producing a 9 kg loss takes lean tissue with it, so the honest question is whether these drugs take a larger share than an equivalent loss by other means. Answering that needs a trial designed around it, and the trials were designed around weight. The same design gap runs through the escalation data, where the reported outcomes are the ones the protocol happened to collect.
What follows from that
Nothing here supports a claim that these drugs spare muscle, and nothing here supports a claim that they destroy it. What it supports is measuring: resistance training and protein intake are the standard advice for preserving lean tissue during any weight loss, and they are advice a prescriber gives rather than a product a seller sells. Whether a service includes that kind of support, or sells medication alone, is recorded per seller in the bundled-care reviews and across the full set. What none of them can offer is a figure for how much muscle you will lose, because that figure has not been established.