Weight Regain After Stopping GLP-1s: What the Withdrawal Data Shows

GLP-1 receptor agonists reduce body weight primarily by acting on appetite. They amplify satiety signaling in the hypothalamus and brainstem, slow gastric emptying, and lower the reward salience of food, which together drive a sustained reduction in energy intake. Dual and triple agonists such as tirzepatide (GLP-1 plus GIP, glucose-dependent insulinotropic polypeptide) and retatrutide (GLP-1 plus GIP plus glucagon) add further metabolic actions.

The key mechanistic point for discontinuation is that these are receptor-occupancy effects, not a structural reset of the body's weight set point. When circulating drug falls below the level needed to occupy the receptor, appetite suppression fades. Compounding this, weight loss itself triggers well-documented counter-regulatory adaptations: circulating leptin falls, ghrelin rises, and resting energy expenditure declines below what body size alone would predict. During active treatment the drug overrides these pressures. Once it is removed, the counter-regulatory drive is unopposed, and energy intake tends to rise above the reduced maintenance level. This is the physiological basis for what is popularly called GLP-1 rebound weight gain.

The STEP 1 extension (Wilding et al., Diabetes Obes Metab 2022) followed participants from the landmark STEP 1 semaglutide trial for one year after treatment and lifestyle intervention were withdrawn at week 68.

The published trial reported that participants who had reached a mean 17.3% body weight reduction on semaglutide 2.4 mg regained 11.6 percentage points over the following 52 weeks, leaving a net loss of about 5.6% from baseline at week 120. That is roughly two-thirds of the lost weight returning within a year of stopping. The trajectory was gradual and continuous rather than a sudden bounce, consistent with the slow washout of a long-acting agonist and the progressive re-emergence of appetite.

Cardiometabolic variables that had improved on treatment, including systolic blood pressure, lipids, HbA1c, and C-reactive protein, moved back toward their baseline values in parallel with the weight. The authors framed the study as evidence that continued treatment is required to maintain the improvements, which is the standard interpretation of a chronic-disease pharmacotherapy being removed.

STEP 4 (Rubino et al., JAMA 2021) is the cleanest randomized test of the withdrawal effect for semaglutide because it randomized after a run-in. All participants took semaglutide for a 20-week lead-in and lost weight, then were randomized either to continue semaglutide or to switch to placebo for a further 48 weeks, both with lifestyle support.

The divergence was large. From randomization at week 20 to week 68, the continued-semaglutide group lost a further 7.9% of body weight, while the placebo-switch group regained 6.9%. The estimated treatment difference was approximately 14.8 percentage points. Because both arms entered the randomized phase at the same weight, this design isolates the drug effect from the initial loss and from lifestyle factors, and it shows that the regain on stopping is not merely a return of willpower problems but the loss of an active pharmacological signal.

SURMOUNT-4 (Aronne et al., JAMA 2024) applied the same randomized-withdrawal design to tirzepatide. Participants took tirzepatide through a 36-week open-label lead-in, achieving a mean 20.9% weight reduction, then were randomized to continue tirzepatide or switch to placebo for 52 more weeks.

The published trial reported that the placebo-switch group regained a mean 14.0% of body weight from randomization to week 88, while the continued-tirzepatide group lost an additional 5.5%. Expressed as total reduction from the original baseline, that left the continued group at about 25.3% and the withdrawal group at about 9.9%. The trial also reported that the regain in the placebo-switch group was accompanied by a reversal of the cardiometabolic improvements seen during the lead-in. The pattern mirrors the semaglutide data: deeper initial loss from the more potent dual agonist did not protect against regain once the compound was removed.

The three randomized datasets tell a consistent story despite different compounds and designs. The table below summarizes the reported figures.

Two generalizations are supported by the data. First, regain is directional and consistent: every published withdrawal arm gained weight relative to a continuation arm. Second, the amount regained scales roughly with how much was lost, so more effective compounds do not appear to produce more durable loss after they are stopped. Retatrutide, the triple agonist that reported approximately 24% mean loss at 48 weeks in its Phase 2 trial, has no published randomized withdrawal trial to date, so its post-discontinuation trajectory is unknown and should not be assumed to differ.

A frequently overlooked point in the withdrawal literature is that the metabolic benefits regress alongside the weight. In the STEP 1 extension, blood pressure, lipid fractions, HbA1c, and inflammatory markers all drifted back toward baseline as weight returned. The SURMOUNT-4 trial reported the same coupling for tirzepatide: the placebo-switch group's regain was accompanied by a reversal of the blood pressure, glucose, and lipid improvements achieved during the lead-in.

This matters for how the research is interpreted. The weight number is a visible proxy, but the trials suggest that the underlying cardiometabolic risk profile is what tracks the drug. Studying maintenance is therefore not only about the scale reading; it is about whether the metabolic state achieved during treatment can be preserved by any means once the agonist is withdrawn.

The most common practical question - whether slowly tapering off, rather than stopping abruptly, blunts the regain - is not answered by any published randomized trial. The withdrawal arms in STEP 1 extension, STEP 4, and SURMOUNT-4 were fixed-duration or abrupt stops, not structured down-titrations, and none compared a taper against a hard stop as a randomized variable.

What the mechanism predicts is that any schedule leaving the receptor unoccupied will eventually expose the same counter-regulatory drive, so a taper might smooth the timing of regain without changing the endpoint. But that is an inference, not a finding. Approaches studied in the broader obesity literature for maintaining loss after any intervention, including sustained resistance training, high-protein intake, and continued structured monitoring, have their own evidence base but have not been tested specifically as a way to hold weight after GLP-1 withdrawal in a randomized trial. Anyone designing a research protocol around maintenance should treat the taper question as open.

These trials are informative but bounded, and honest framing requires stating the limits.

They cannot tell you that a taper works. No randomized comparison of tapering versus abrupt discontinuation has been published for any GLP-1 agonist.

They cannot tell you retatrutide's withdrawal behavior. The regain figures above come from semaglutide and tirzepatide. Retatrutide has Phase 2 efficacy data but no published discontinuation trial, so its post-stop trajectory is genuinely unknown.

They cannot tell you about individuals. The reported regain figures are group means with wide individual variation; some participants held their loss better than others, and the trials were not designed to identify who or why.

They cannot separate diet and activity contributions cleanly. Lifestyle support continued in the withdrawal arms, so the regain occurred despite ongoing behavioral intervention, but the trials cannot quantify how much a different behavioral program might have changed the result.

And none of this is clinical guidance. These are research findings about how a class of compounds behaves when removed, not instructions for use in humans.

The compounds discussed here are supplied strictly for research purposes only (RUO) and are not for human consumption. See full specifications for Retatrutide, the triple incretin receptor agonist referenced throughout this review.

Related research reviews on this site: