Three arms, three sets of numbers, and they are not on the same scale. Aerobic exercise without a structured deficit produced a mean weight reduction of about 2.8 kg at 150 minutes a week and 4.2 kg at 300 minutes a week across 116 randomised trials (Jayedi 2024). GLP-1 medication trials report mean weight changes of −14.9% and −20.9% of body weight (Wilding 2021; Jastreboff 2022). Exercise added to a calorie-restricted diet preserved 0.87 kg more fat-free mass than dieting alone — roughly 46% of what would otherwise be lost (Deller 2026). And in the one randomised trial that tested all three arms against each other, the combination beat exercise but did not significantly beat medication alone (Lundgren 2021).
This article sets out each of those figures with its population, its design and its limits. Individual results vary in every trial described.
Fat-free mass: everything in the body that is not fat — muscle, bone, organs, water and glycogen. Most trials measure this rather than muscle, because a scan estimates it directly. When a study says "fat-free mass", writing "muscle" instead overstates what was measured.
What exercise alone does to body weight
The largest dose-response synthesis pooled 116 randomised trials in 6,880 adults with overweight or obesity — 61% female, mean age 46. Per 30 minutes a week of aerobic exercise, it found body weight −0.52 kg (95% CI −0.61 to −0.44, moderate certainty), waist circumference −0.56 cm (high certainty), body fat percentage −0.37% (moderate), visceral adipose tissue −1.60 cm² (high) and subcutaneous adipose tissue −1.37 cm² (moderate).
At the doses people actually plan around: 150 minutes a week produced −2.79 kg (95% CI −3.29 to −2.29) and −2.08% body fat; 300 minutes a week produced −4.19 kg (95% CI −5.98 to −2.41), with waist circumference falling between 4.21 and 5.34 cm depending on intensity. Reductions were linear or monotonic up to 300 minutes a week (Jayedi 2024).
One qualification changes how that should be read. Calorie restriction was a co-intervention in some of the included trials, handled by the authors in a post hoc subgroup analysis rather than by exclusion. So this is the best available estimate of what adding aerobic exercise does, not a clean measurement of exercise in isolation. And no fat-free-mass outcome is reported anywhere in it.
What exercise does to fat-free mass in a deficit
This is where the training evidence is strongest, and it is a different question from weight.
A 2026 systematic review with pairwise and network meta-analysis pooled 34 randomised trials in 1,455 people with overweight or obesity. Adding exercise to calorie restriction preserved 0.87 kg more fat-free mass than calorie restriction alone (95% CI +0.59 to +1.16, p≤0.001). The authors' own summary sentence: "On average, EX prevented nearly half of FFM loss (45.7%)" (Deller 2026).
A separate network meta-analysis of 62 randomised trials in 4,429 healthy participants approaches it from the other side and reaches a compatible answer. Ranking nine intervention arms against a no-intervention control for lean body mass preservation, calorie restriction alone was the only arm whose confidence interval excluded zero: −1.66 (95% CI −3.12 to −0.19). Every diet-plus-exercise arm's interval crossed zero — meaning none of them differed significantly from doing nothing at all (Xie 2025).
Two older syntheses fill in the picture. Across 52 studies in men and postmenopausal women aged 50 and over, 81% of the energy-restriction-only groups lost 15% or more of their body weight as fat-free mass, against 39% of the groups that also exercised (Weinheimer 2010). And across 114 trials in 4,184 people, lean mass was statistically unchanged when resistance training accompanied caloric restriction (effect size around −0.3 kg, p=0.550 to 0.727) (Lopez 2022).
Note two limits that run through all of it. The outcome is fat-free mass, not muscle: only two of Deller's 34 trials measured skeletal muscle mass specifically. And the senior author of that 2026 analysis declares lecture, consultation and honorarium payments from several pharmaceutical companies including the manufacturer of a GLP-1 medication — disclosed, pre-registered, and worth stating rather than leaving out.
Which training mode — and what the comparison does not show
The temptation here is to declare weights the winner and cardio the loser. The data do not support that.
In the network analysis, mixed training preserved +1.20 kg of fat-free mass (95% CI 0.67 to 1.73, p<0.001), strength training +0.83 kg (0.17 to 1.49, p=0.013), and endurance training +0.51 kg (−0.04 to 1.05, p=0.067) — the last of these falling just short of the significance threshold. But the paper states plainly that subgroup testing found no significant differences between training modes (Deller 2026). The 62-trial network analysis says the same in a different shape: every exercise arm's interval overlaps every other one (Xie 2025).
Missing p<0.05 is an absence of proof, not proof of inferiority. The defensible line is that combined and resistance training are the modes with demonstrated fat-free-mass protection, and that aerobic training alone has not been shown to protect it. The indefensible line is that aerobic exercise is worse for muscle, which no pooled comparison supports.
One randomised trial does show a mode gap directly, and its population is narrow. In 160 previously sedentary obese adults with a mean age around 70, over six months of dieting, lean mass fell about 5% with aerobic exercise alone, 2% with resistance training alone and 3% with both — while combined training produced the largest gain in physical function (Villareal 2017). That trial had no diet-only arm, so it compares modes against each other rather than against dieting without exercise. We go into the aerobic question in more depth in cardio only and muscle loss.
What medication alone achieved in trials
In STEP 1, semaglutide 2.4 mg weekly produced a mean weight change of −14.9% at 68 weeks against −2.4% with placebo (Wilding 2021). In SURMOUNT-1, tirzepatide 15 mg weekly produced −20.9% at 72 weeks against −3.1% with placebo, on the treatment-regimen estimand (Jastreboff 2022). Individual results vary; these describe trial populations under supervised conditions.
Routine practice produces smaller numbers. In a US cohort of 7,881 adults with overweight or obesity without diabetes, mean weight reduction at one year was 8.7%, with 80.8% of patients on a lower-than-maximal maintenance dose (Gasoyan 2025).
Those medication figures are percentages of body weight over 68 to 72 weeks; the exercise figures above are kilograms over trials of varying length in different populations. Lining them up in a single sentence would be arithmetic rather than evidence. The only way to compare the arms properly is a trial that randomised people to both, in the same units, at the same time — which brings us to the next section.
The one trial that randomised all three
After an eight-week 800 kcal/day low-calorie diet that produced a mean 13.1 kg loss, 195 adults with obesity (BMI 32 to 43, without diabetes) who had lost at least 5% were randomised for one year to four arms: exercise plus placebo; GLP-1 medication plus usual activity; exercise plus medication; or placebo plus usual activity (Lundgren 2021).
At one year, against placebo:
| Arm | Weight difference vs placebo | 95% CI | P |
|---|---|---|---|
| Exercise | −4.1 kg | −7.8 to −0.4 | 0.03 |
| Medication | −6.8 kg | −10.4 to −3.1 | <0.001 |
| Both | −9.5 kg | −13.1 to −5.9 | <0.001 |
The head-to-head comparisons are where the honesty lives. The combination beat exercise alone by −5.4 kg (95% CI −9.0 to −1.7, P=0.004). It did not significantly beat medication alone: −2.7 kg (95% CI −6.3 to 0.8, P=0.13).
On body composition, the combination reduced body-fat percentage by 3.9 percentage points, roughly twice the reduction in the exercise group (−1.7 points, P=0.02) and the medication group (−1.9 points, P=0.009). Only the combination arm improved cardiorespiratory fitness. Increased heart rate and gallstones were seen more often in the medication-only group than in the combination group.
Four constraints on reading this. The medication was liraglutide 3.0 mg daily — a different molecule and a different dosing schedule from the weekly medicines in the STEP and SURMOUNT trials, and its results should not be transferred to them. The trial reports body-fat percentage, not lean mass, and a falling fat percentage can arise from fat loss alone. It ran in 195 people in one country as a one-year maintenance phase after a very-low-calorie run-in. And the P=0.13 result is the part most summaries drop.
The Singapore gap
Singapore's national guidelines ask adults for 150 to 300 minutes a week of mostly aerobic moderate-intensity activity plus muscle-strengthening work on at least two days a week, at 8 to 12 repetitions per set (SPAG 2022).
The same guidebook reproduces National Population Health Survey 2020 figures: 84.7% of residents aged 18 to 74 met the physical activity guideline in 2024, while only 35.7% did sufficient muscle-strengthening activity — 44.1% at ages 18 to 29, 25.5% at 60 to 74, and 40.1% of men against 27.8% of women. Those are the survey's numbers. The reading of them is ours: the aerobic half of the guideline is broadly being met and the strength half is being skipped, in a country where the same guideline asks for both. Our practical version of the strength half is in minimum effective resistance training.
How to read the maths
Four statements the evidence supports, stated at the size the evidence supports them:
- Exercise without a structured deficit moves weight modestly — about 2.8 kg at 150 minutes a week, about 4.2 kg at 300 (Jayedi 2024).
- Exercise added to a deficit protects fat-free mass — about 0.87 kg more preserved, roughly 46% of what would otherwise go (Deller 2026).
- Dieting without exercise is the one arm that reliably loses lean mass in the network comparison (Xie 2025).
- In the only randomised head-to-head, exercise, medication and the combination all beat placebo, and the combination's advantage over medication alone did not reach significance (Lundgren 2021).
At GetLean, our philosophy is that the medication is a catalyst and the training is what determines what a weight loss is made of. That is our position on how to use the evidence above, stated as ours. Which of these arms is appropriate for any individual is a clinical decision, and you can check your eligibility to have that conversation with a doctor.
Common questions
How much weight does exercise alone take off?
Across 116 randomised trials in 6,880 adults with overweight or obesity, aerobic exercise produced a mean weight reduction of 2.79 kg at 150 minutes a week (95% CI −3.29 to −2.29) and 4.19 kg at 300 minutes a week (95% CI −5.98 to −2.41), with a graded response up to 300 minutes (Jayedi 2024). Calorie restriction was a co-intervention in some of the included trials, so this is not a clean exercise-only figure.
Does exercise protect muscle while dieting?
It protects fat-free mass, which is a broader measure. Pooling 34 randomised trials in 1,455 people, adding exercise to a calorie-restricted diet preserved 0.87 kg more fat-free mass than dieting alone (95% CI +0.59 to +1.16, p≤0.001) — the authors' summary being that exercise prevented nearly half of the fat-free-mass loss, 45.7% (Deller 2026). Only two of the 34 trials measured skeletal muscle mass specifically.
Is cardio worse than weights for keeping muscle?
Not on the direct comparison. In the network analysis, mixed training preserved 1.20 kg of fat-free mass and strength training 0.83 kg, both statistically significant, while endurance training preserved 0.51 kg and missed significance at p=0.067 — but the paper states that subgroup testing found no significant differences between training modes (Deller 2026). Missing a significance threshold is an absence of proof, not proof of inferiority.
Has anyone randomised people to exercise, medication or both?
Once. After an eight-week low-calorie diet, 195 adults with obesity were randomised for a year to exercise plus placebo, GLP-1 medication plus usual activity, both combined, or placebo. Versus placebo, weight change was −4.1 kg with exercise, −6.8 kg with medication and −9.5 kg with both. The combination beat exercise significantly (−5.4 kg, P=0.004) but did not significantly beat medication alone (−2.7 kg, P=0.13) (Lundgren 2021). The medication was liraglutide 3.0 mg daily.
How many Singaporeans do enough strength training?
About a third. National Population Health Survey 2020 figures reproduced in Singapore's physical activity guidelines show 76.4% of adults aged 18 to 74 met the physical activity guidelines while only 33.8% did sufficient muscle-strengthening activity — 44.1% at ages 18 to 29 and 25.5% at 60 to 74 (SPAG 2022).