Exercise plus a GLP-1: what the randomised trials show, and what nobody has tested yet

Most of what this site says about training is inferred from weight loss in general. One trial tested exercise and a GLP-1 head to head and in combination, under randomisation, for a year. It deserves reading in full.

S-LiTE: exercise, liraglutide, or both, after a diet

Design: adults with obesity (BMI 32 to 43) without diabetes followed an 8-week low-calorie diet. The 195 who completed it, having lost a mean 13.1 kg, were randomised to one of four one-year strategies: a supervised moderate-to-vigorous exercise programme plus placebo; liraglutide 3.0 mg daily plus usual activity; exercise plus liraglutide; or placebo plus usual activity (PubMed 33951361).

Weight at one year, difference from placebo: exercise -4.1 kg (95% CI -7.8 to -0.4); liraglutide -6.8 kg (95% CI -10.4 to -3.1); combination -9.5 kg (95% CI -13.1 to -5.9). The placebo group regained about half the diet-induced loss; the combination group kept losing.

Body composition: the combination reduced body-fat percentage by about twice as much as either exercise or liraglutide alone, and the combination group was the one that improved cardiorespiratory fitness and glycaemic measures together.

Why it matters here:

1. The effects were roughly additive. Exercise added to the drug rather than being made redundant by it. Nothing in the drug's mechanism replaces what training does.
2. Composition improved most with both. The body-fat percentage result shows exercise changing what the loss is made of, not only how much of it there is. That is the concern raised by the DXA substudies, addressed in a randomised setting.
3. The exercise was real. Two supervised group sessions a week of vigorous intensity plus individual sessions, targeting about 150 minutes of moderate or 75 of vigorous activity weekly, with heart-rate monitoring. Not a leaflet.

Caveats: liraglutide is a daily GLP-1 with smaller weight effects than semaglutide or tirzepatide; the population was younger and without diabetes; the exercise was supervised; and the trial started after a diet phase, so it is a maintenance trial more than a loss trial. The direction of the findings is the point, not the exact kilograms.

Villareal 2017: what kind of exercise during weight loss

Design: 160 adults aged 65 and over with obesity were randomised to a weight-management programme plus aerobic training, resistance training, or both, or to a control group with neither, for six months. The primary outcome was a physical performance test (PubMed 28514618).

Result: all three exercise groups lost about 9 percent of body weight. The combination group improved physical performance most (21 percent versus 14 percent in each single-mode group). Lean mass and bone mineral density fell less in the resistance and combined groups than in the aerobic-only group. Aerobic capacity improved most in the aerobic and combined groups.

Why it matters here: this is the cleanest demonstration that the type of exercise during weight loss determines what the loss is made of. Aerobic exercise alone did not protect lean mass or bone. Adding resistance training did, without costing weight loss. For anyone over 60 on a GLP-1, where lean loss matters most, it is the central piece of evidence behind the advice to lift.

What has not been tested

• No published large randomised trial has assigned people on semaglutide or tirzepatide to a resistance-training programme versus none and measured body composition. Trials are under way; until they report, the inference runs from S-LiTE, Villareal and the general weight-loss literature.
• No trial has compared exercise doses on a GLP-1, so the three-times-weekly strength recommendation in the 2025 joint advisory (PubMed 40445127) is a consensus judgement built on the above rather than a dose-finding result.
• The Neeland review notes drug candidates aimed at preserving muscle during GLP-1 therapy in development; none is approved, and none should change what anyone does now (PubMed 38937282).

What follows from the evidence

Train while you lose, not after. Lift, because aerobic work alone does not hold lean mass. Keep it up when the drug's effect plateaus and if the drug ever stops, because S-LiTE is the best evidence that exercise carries part of the maintenance load (maintenance after stopping). And eat the protein that makes the training count (protein targets). The specifics are in resistance training basics and aerobic activity and steps.

The trials above used brand products (liraglutide as Saxenda in S-LiTE) at label doses under supervision. Compounded GLP-1 products are not FDA approved and are not interchangeable with brand products; the exercise physiology does not change with the product, but the trial numbers were not measured on them.