What Happens When You Stop a GLP-1?

Most of what you read about GLP-1 receptor agonists concerns starting them, which is understandable, though the research question that has been answered most cleanly is the opposite one. Between 2021 and 2024, three large randomized trials deliberately took the drug away from people who were doing well on it and then measured what happened next.

That design, called a randomized withdrawal trial, is unusual and unusually informative, because it removes the confound that ruins observational data, where the people who stop a medication are often the people it was not working for in the first place. In these trials, participants had already lost substantial weight before anyone was assigned to placebo, so whatever happened afterward is attributable to the withdrawal itself rather than to who chose to quit.

What follows covers what each trial measured, what the regain curves actually look like, why your physiology pushes in that direction, and where the evidence genuinely runs out. This is educational content about published research, and nobody should start, stop, taper or change a prescribed medication on the strength of an article.

Yes, in every randomized withdrawal trial conducted so far, and the regain begins sooner than most people expect it to. Across STEP 1, STEP 4 and SURMOUNT-4, the weight curve reverses within weeks of the last dose and keeps climbing for at least a year afterward.

The nuance worth holding onto is that regain has been partial rather than total in every trial that ran long enough to observe it, so participants generally end up somewhere between their starting weight and their lowest weight rather than back where they began. How far up that curve you travel appears to depend on how much you lost, how long you were treated, and what happened to your eating and activity patterns once the drug cleared your system.

The STEP 1 extension followed 327 participants for one year after the 68 week parent trial ended, with no continued study drug and no ongoing lifestyle support, which makes it the longest published look at what semaglutide withdrawal does over time.

At week 68 the semaglutide group had lost a mean of 17.3 percent of body weight, and by week 120, a full year after the last injection, that figure had shrunk to 5.6 percent below baseline. The placebo group, which had lost 2.0 percent by week 68, sat at 0.1 percent below baseline at week 120. Put plainly, the semaglutide group handed back about 11.6 percentage points of the 17.3 they had lost, which works out to roughly two thirds of it.

The metabolic markers moved along with the weight rather than staying put. HbA1c, fasting lipids, systolic blood pressure and C-reactive protein had all improved during treatment, and all of them drifted back toward their starting values across the year off. The trial did not show these markers becoming worse than baseline, so what it documented was the improvement being returned rather than any kind of overshoot into worse territory.

Regain typically becomes measurable within four to eight weeks of your last dose, which lines up with the pharmacokinetics of the drugs involved. Semaglutide has a half life of about one week, so plasma levels fall to a negligible fraction of steady state across roughly five weeks, while tirzepatide clears somewhat faster with a half life near five days.

STEP 4 gives the clearest picture of that early period because its follow up was tightly scheduled. Participants ran in on semaglutide 2.4 mg for 20 weeks and lost a mean of 10.6 percent, and they were then randomized either to continue the drug or to switch to placebo for another 48 weeks. Those switched to placebo gained 6.9 percent of body weight over that period while the continuers lost an additional 7.9 percent, an estimated treatment difference of 14.8 percentage points.

What makes STEP 4 so useful is that both arms had already succeeded before anyone was randomized, so the divergence afterward reads cleanly as what the ongoing drug exposure was contributing rather than as a comparison between people who responded and people who never did.

The pattern looks much the same, with the caveat that SURMOUNT-4 participants had further to fall because they had lost considerably more to begin with.

SURMOUNT-4 randomized 670 participants after a 36 week open label tirzepatide lead in, during which they lost a mean of 20.9 percent of body weight. Over the following 52 weeks, those switched to placebo regained 14.0 percent while those who continued tirzepatide lost a further 5.5 percent, so at the end of the full 88 weeks the continuation group sat about 25.3 percent below their original baseline and the withdrawal group about 9.9 percent below it.

Sitting almost ten percent below your starting weight a year after the last dose is a meaningful result, and it exceeds what most interventions without medication produce at all. Measured against the 20.9 percent they had reached, though, more than two thirds of the loss had been given back, which lands strikingly close to the STEP 1 extension figure despite a different molecule, a different schedule and a different trial population.

Your physiological adaptations to weight loss outlast the weight loss itself, and GLP-1 drugs suppress those adaptations rather than reversing them, which is the whole explanation compressed into one sentence.

None of this was introduced by GLP-1 drugs, and it was described carefully well before they existed. In a 2011 New England Journal of Medicine study, Sumithran and colleagues put participants through a ten week very low energy diet and then measured appetite regulating hormones at 62 weeks, finding that circulating ghrelin remained elevated above the levels seen before dieting, leptin remained suppressed, and subjective hunger remained higher than at baseline a full year after the weight loss had ended.

Layered on top of that hormonal picture is adaptive thermogenesis, which is the tendency for resting energy expenditure to fall further than the change in body composition alone would predict. A body that has been reduced in weight burns fewer calories than a body of the same size that was never reduced, and that gap does not close on its own with time.

A GLP-1 receptor agonist works partly by slowing gastric emptying and partly through central appetite signalling in your hypothalamus and hindbrain, so while the drug is present it offsets the hunger signal your body is generating. Once it clears, that underlying signal is still there, still elevated, and still pointing at your old weight, which means regain in these trials is better understood as an unmasking than as a rebound.

This is where the evidence thins considerably, and it deserves rather more caution than it usually receives online.

Weight lost on GLP-1 receptor agonists includes both fat mass and lean mass, and body composition substudies within the semaglutide and tirzepatide programmes have reported lean mass accounting for a meaningful fraction of the total, with figures commonly cited between 25 and 40 percent depending on the population studied and the measurement method used. That proportion is broadly comparable to what caloric restriction produces without any medication at all, so it is a feature of losing weight quickly rather than something these drugs uniquely cause.

The concern raised in the literature is that regained weight tends to be preferentially fat mass while lost lean mass is not automatically restored, which would imply a net worsening of body composition across a cycle of loss and regain. The mechanism is plausible and the pattern has been described in the older weight cycling literature, yet what does not exist is a well powered randomized body composition analysis measuring lean and fat mass before, during and after GLP-1 discontinuation. Anyone stating this as established fact is running ahead of the data.

Resistance training and adequate protein intake are the standard research supported levers for preserving lean mass during any energy deficit, and both are being studied in the GLP-1 context, though the trials aimed specifically at the discontinuation period have not been run yet.

There is no randomized evidence that tapering helps, because every trial discussed above used abrupt discontinuation and the comparison of interest has simply never been tested head to head.

Tapering is nonetheless widely discussed and sometimes practised, and the rationale behind it is not unreasonable, since a gradual reduction in receptor agonism might let appetite regulation settle again more slowly and give behavioural strategies time to take over. That remains a hypothesis rather than a finding, and it is worth labelling as one.

The better supported alternative is maintenance dosing, where a lower dose continues indefinitely instead of stopping, and its logic comes straight out of the trial data, because the continuation arms in STEP 4 and SURMOUNT-4 did not merely hold their weight but carried on losing. Whether a reduced maintenance dose captures most of that benefit is an active research question rather than something anybody has settled.

Far fewer people stay on them than the trial literature would lead you to expect, and that gap matters for how you read the discontinuation data.

Clinical trial participants receive free medication, structured follow up and support through dose escalation, none of which describes ordinary practice. Real world pharmacy claims analyses have consistently reported much lower persistence, with a large fraction of people prescribed a GLP-1 for weight management no longer filling prescriptions within the first year, and cost, insurance coverage changes, supply interruptions and gastrointestinal side effects during titration all contribute to that attrition.

The practical implication is that discontinuation is not an edge case worth setting aside. For a substantial share of the people who start these medications, the withdrawal trials describe the most likely eventual scenario rather than a distant hypothetical.

The withdrawal data are the strongest single argument for treating obesity as a chronic and relapsing condition rather than an acute one, which is how major clinical bodies have described it for years.

The comparison usually drawn is to blood pressure medication, which lowers blood pressure while it is being taken and stops lowering it once you stop taking it, and nobody treats that as a failure of the drug. The STEP and SURMOUNT withdrawal arms show GLP-1 receptor agonists behaving in exactly the same way, and the fact that this surprised so many people says more about how weight is culturally understood than about the pharmacology involved.

That framing has limits worth naming honestly, because it does not tell you whether indefinite treatment is right for you, it does not address what any of this costs, and it does not resolve the open questions about very long term exposure. What it does accomplish is making the trial results legible, since the drug was doing something continuous and stopping it stopped that thing.

Three randomized withdrawal trials point in the same direction with unusual consistency, showing that stopping a GLP-1 receptor agonist reverses a large fraction of the weight loss it produced, though not all of it, and that the metabolic improvements track the weight rather than persisting on their own. The underlying reason is that these drugs suppress your body's defence of its previous weight without abolishing it, so removing the drug reveals a hunger signal that never actually left.

What remains genuinely unknown is more interesting than what has been settled, including whether tapering helps, what an optimal maintenance dose looks like, what a cycle of loss and regain does to body composition over time, and which individuals hold onto their loss without ongoing treatment. Those are open research questions, and the honest answer to most of them today is that the trial has not been done.

This article is educational content about published clinical research rather than medical advice, and it is not a recommendation to start, continue, taper or discontinue any medication. Decisions about prescription therapy belong with a qualified healthcare provider who knows your history.