For someone training seriously, the interesting question about these drugs is not how much weight comes off. It is what fraction of it is tissue they spent years building, whether the gym offsets any of it, whether anything measurable happens to performance, and whether a drug sold for weight loss is allowed in competition. Those are four separate literatures of four very different sizes, and the pooled kilogram figures behind the first one are in the body-composition article.
The number flips sign depending on the denominator
A substudy of a randomized trial scanned 178 adults with type 2 diabetes by dual-energy X-ray absorptiometry over 52 weeks on semaglutide 1.0 mg or canagliflozin. On semaglutide, total fat mass fell 3.4 kg from a baseline of 33.2 kg, and total lean mass fell 2.3 kg from a baseline of 51.3 kg. Lean mass as a proportion of body mass rose 1.2 percentage points, from 59.4%.[1]
Both of those come from the same scan. An absolute loss of lean tissue and an improvement in body composition are not contradictory findings; they are the same finding read against two denominators. Which one a marketing page quotes tells a reader almost nothing about the drug and quite a lot about the page. The substudy had no placebo arm, and its authors call any isolated drug effect speculative.[1]
The lean fraction, and the comparison that matters
A 2026 meta-analysis restricted itself to randomized trials measuring body composition by DXA or MRI, and reached 20 trials covering 15,782 participants. Lean mass as a proportion of total weight lost came to 35.2% on semaglutide (95% CI, 31.5 to 38.9), 25.4% on tirzepatide (95% CI, 22.8 to 28.0) and 26.8% on liraglutide (95% CI, 23.1 to 30.5).[2]
Two things in that row are worth stopping on. The semaglutide and tirzepatide intervals do not overlap, and the agent that produces the larger weight loss has the lower lean fraction — the opposite of the intuition that a bigger effect must cost more muscle. Then the comparator: intensive lifestyle intervention alone came to 26.2% (95% CI, 24.1 to 28.3), statistically indistinguishable from the drugs at p = 0.42.[2] On this evidence the drug is not what takes the lean tissue. Losing weight is.
One arm did move the ratio. Lifestyle plus resistance training produced a lean fraction of 17.5% (95% CI, 14.2 to 20.8), an interval that does not overlap any of the drug arms.[2] That is an indirect comparison across trials rather than a randomized head-to-head against a GLP-1, and the trials that randomized training alongside one of these drugs are covered in the training article.
A narrative review in the same year put the arithmetic in terms a lifter recognizes. It describes lean mass losses on incretin therapy of roughly 6 kg, comparable to a decade or more of aging, against supervised resistance programs running longer than ten weeks that produce increases of about 3 kg in lean mass and about 25% in strength.[3] Those are the authors’ synthesized figures rather than a pooled estimate, and they are the clearest statement available of what training is being asked to offset.
The instrument decides the size of the problem
A 2026 systematic review screened 8,102 titles and analyzed 35 primary studies, with a median duration of 26 weeks and a median of 78 participants. Measured by bioelectrical impedance or DXA, the median share of weight loss appearing as a muscle-based index was about 29%. Measured by CT or MRI it was about 25.3%.[4] The smaller number is the more alarming one, because the review’s prespecified benchmark for CT and MRI is 15% against 25% for the impedance and DXA methods. The same phenomenon clears one threshold and badly misses the other.
There is a more basic reason DXA overstates this. Fat-free mass is not muscle: it contains roughly 55% skeletal muscle alongside organs, bone and water, so a headline kilogram of fat-free mass is closer to half a kilogram of muscle.[5] A companion analysis from the same group adds the obligatory part — adipose tissue itself contains water and protein that a scan counts as fat-free mass, so some fat-free mass loss is an arithmetic consequence of losing fat and is not muscle at all.[6]
Which is why two 2026 meta-analyses of overlapping evidence reached opposite verdicts. One concluded that lean mass loss during weight reduction is substantial; another, drawing on 36 studies with 24 in its pooled analysis, concluded that these drugs achieve weight loss primarily through selective fat-mass reduction with relative preservation of lean tissue.[2][7] Neither is fabricating. They weighted different studies, different instruments and different durations, and a page quoting one without the other is presenting a settled question that is not settled.
Strength and mass do not move together
The most useful single study here followed 106 patients with obesity on semaglutide 2.4 mg for a year, with DXA, handgrip dynamometry and resting energy expenditure at baseline, seven months and twelve. Weight fell 13% and total fat mass 18%. Lean mass fell 3 kg by month seven and then stabilized. Over the same year handgrip strength rose 4.5 kg, and the prevalence of sarcopenic obesity fell from 49% to 33%.[8]
Three instruments, three directions: the scan says muscle was lost, the dynamometer says the patient got stronger, and the diagnostic criteria say a third of the sarcopenic patients stopped meeting them. That study was single-arm with no control group, so some of the grip improvement may be practice rather than physiology. It still demonstrates that a lean-mass number cannot be read as a strength number.
A 2026 review states the general case directly: lean soft tissue loss is not a reliable predictor of strength change on these drugs. It also names the split. Short and mid-term trials in adults with obesity show handgrip strength statistically preserved despite reductions in lean soft tissue, while longitudinal work in older adults with type 2 diabetes reports falling grip strength and accelerated sarcopenia with prolonged use.[9]
Where the counter-evidence lives
Two studies sit on the unfavorable side and belong in any honest summary. A 24-month retrospective cohort of 220 patients on semaglutide against 212 controls found grip strength that initially improved and then declined in men, while declining throughout in women, with gait speed falling significantly in both. Semaglutide dosage was an independent predictor of muscle loss.[10] A twelve-month comparative study of 141 older men with type 2 diabetes found significant reductions in grip strength, gait speed, appendicular skeletal muscle index and short physical performance battery scores on semaglutide against sitagliptin, alongside larger rises in two circulating markers of neuromuscular junction and neuronal damage.[11]
Both are older adults with diabetes, neither is randomized, and neither describes a 30-year-old under a barbell. What the second one adds is a mechanism a scan cannot see: if the change is at the neuromuscular junction rather than in muscle volume, a DXA-only analysis is blind to it. The age dimension of all this is in the older-adults article.
Performance has barely been measured at all
The 2026 systematic review of 35 studies states flatly that no study reported objective physical function outcomes.[4] A narrative review of incretin therapy in sport describes preclinical work showing enhanced exercise endurance, mitochondrial biogenesis and glucose uptake, and then says that clinical trials show no consistent improvement in physical performance in humans.[12] Animal data and human data point in opposite directions, and only one of them is about people.
No randomized trial has enrolled athletes or resistance-trained people at all. Every figure on this page comes from cohorts with obesity, type 2 diabetes or heart failure. Anyone extrapolating from a 46 kg/m² cohort to a 12% body-fat lifter is doing that without evidence, and the protein advice usually attached to it is examined in the diet article.
The anti-doping position, which is two facts
No GLP-1 receptor agonist appears on the World Anti-Doping Agency’s 2026 Prohibited List. Semaglutide, tirzepatide, liraglutide, exenatide and dulaglutide are absent from it, and the incretin class is absent from section S2 on peptide hormones and from section S4 on hormone and metabolic modulators, whose metabolic entries are insulins and insulin-mimetics, meldonium and trimetazidine.[13]
The second fact is the one the published commentary omits. WADA also publishes a Monitoring Program for substances that are not prohibited but which it wishes to watch for patterns of misuse. Item 6 of the 2026 program reads “Markers of Semaglutide and Tirzepatide: In and Out-of-Competition”.[14] Semaglutide first appeared on that program for 2024 and tirzepatide was added for 2026.[14] Nothing there is a violation. It does mean samples are being examined for these drugs, in and out of competition, and have been since 1 January 2024.
A sport’s own governing body is a separate matter from the Prohibited List, and weight-category sports in particular set their own rules about making weight. Neither this page nor WADA’s list answers that question for a particular federation.
What a competitor is actually deciding
The evidence supports a narrow set of statements. Roughly a quarter to a third of weight lost is lean tissue, which is about what losing weight costs by any method. Resistance training is the only intervention associated with a materially better ratio. Strength does not track mass, and the direction of the strength result depends on how long treatment runs and who is taking it. And performance, in the sense a coach means it, has not been measured.
One further distance applies to anything bought on a subscription. Compounded semaglutide and tirzepatide are not FDA-approved and are not reviewed by the FDA for safety, efficacy or quality before they are dispensed, and every scan above was performed on branded product at labeled doses — a gap set out in the compounding article. For a tested athlete there is a second consequence: a vial whose contents are not independently verified is a vial whose contents cannot be declared with confidence. What the sellers charge for it is recorded on the tirzepatide board, and how this site establishes a figure is described in the methodology.