Stopping a GLP-1 brings back about 60% of lost weight within a year

Why is there no evidence-based GLP-1 taper schedule?

No GLP-1 taper schedule is evidence-based, because nobody has tested one: every randomized study of stopping switched participants from the drug to placebo abruptly, in a single step.

Most people arrive at this subject with a specific request. They want the taper schedule: how many steps, how long at each one, and what the safe way down looks like. That is a reasonable thing to want, and the answer is uncomfortable. Not one of the withdrawal trials tapered.

Anyone who calls their taper schedule evidence-based is describing a practice pattern. A practice pattern is real and useful, but it is not a trial result, and it should not be presented as one.

The evidence answers a better question much more fully: whether you are stopping, or looking for your lowest effective maintenance dose. Those are two different plans with two different bodies of evidence behind them, and one is far better supported than the other. Getting that distinction right is the difference between a decision you understand and a decision that happens to you.

Who needs a different conversation before stopping a GLP-1?

Three groups need a different conversation before stopping a GLP-1: people on insulin or a sulfonylurea, people with an eating disorder history, and people treated for type 2 diabetes.

Insulin or a sulfonylurea: stopping changes your diabetes treatment

If you take insulin or a sulfonylurea, stopping a GLP-1 is not a weight decision. It is a change to your diabetes treatment.

GLP-1 medications lower blood sugar. When they are started, the prescribing information directs the prescriber to consider reducing the dose of any insulin or insulin secretagogue taken alongside them, because the combination raises the risk of hypoglycemia (Wegovy prescribing information, DailyMed, revised June 2026). The sulfonylureas include the tablets glipizide, glyburide, and glimepiride. If you are not sure whether one of your diabetes medications belongs to that family, your pharmacist can tell you in about ten seconds.

That reduction matters on the way out. If your insulin or sulfonylurea dose was lowered while you were on a GLP-1, then by design you are on less diabetes medication than you would otherwise need. Remove the GLP-1 and that reduction is no longer correct.

The risk runs in both directions, and both are real. Left alone, you can lose glycemic control; adjusted without supervision, your blood sugar can drop dangerously low. The person who prescribes your diabetes medication has to manage the change, with a plan for monitoring your blood glucose while it happens.

Managing that change is one of the few solid clinical arguments for a supervised, stepwise discontinuation rather than a sudden one, and no general article can substitute for that supervision.

The timing trap. The weekly medicines leave slowly: semaglutide is still present in the circulation for roughly five to seven weeks after a final dose. Your blood sugar does not jump back the day you stop; it drifts up over weeks. Any diabetes medication increased in the first days after stopping is added on top of a drug that is still working, and that is the sequence that produces a severe low.

The timing is a reason to have the adjustment planned and monitored rather than made reactively. It is also a reason to check your glucose more often through this period, not less.

What a low feels like. Know the signs before you need to: shakiness, sweating, a pounding heart, sudden hunger, confusion, or feeling faint. Treat a low immediately with fast-acting sugar, then tell your prescriber it happened, because it means a dose needs changing.

Lows caused by a sulfonylurea deserve particular respect. They can last for hours and return after the symptoms have eased, so a low on one of those tablets is a reason to be seen, not a reason to eat something and carry on.

A history of an eating disorder: a high-risk window

If you have a history of a restrictive eating disorder, active or in remission, the weeks after stopping are a specific high-risk period, and you should not navigate them alone.

Anorexia, atypical anorexia, and purging behaviors carry a different risk from binge eating disorder, and the difference matters. For a restrictive illness, a year with hunger pharmacologically switched off can have felt like the thing finally working. Appetite returning while weight moves upward is close to a designed relapse trigger. If that is your history, tell whoever knows it that you are coming off, before you come off.

If your history is binge eating disorder, the risk runs the other way. The drive to binge can return, and that return is a treatable clinical event rather than a discipline failure. Binge eating disorder has its own care, separate from the GLP-1.

Type 2 diabetes: a different label and a different decision

If you take a GLP-1 for type 2 diabetes rather than for weight, your situation is different, and the sections below are written about weight management.

The same molecules carry different indications, different approved labels, and different maximum doses depending on which product you were prescribed. For you, stopping has a defined consequence for blood sugar control, and that decision belongs entirely to your prescriber.

What if you have already stopped your GLP-1?

If you already stopped a GLP-1 and the weight is coming back, you did nothing wrong: three randomized trials document that regain, and it is what the pharmacology predicts.

Writing about coming off these medications almost always addresses someone making a considered exit: a plan, a prescriber, a date on the calendar. That person exists, and they are far from the only person reading.

How many people stay on tirzepatide in real-world use?

In a real-world cohort of 755 adults taking tirzepatide without type 2 diabetes, persistence at six months was 54.2 percent. Barely half were still taking it half a year in. Among those who continued, 91.1 percent were still at a middle dose or lower by their fifth refill, far slower escalation than the trials used (Hunter Gibble et al., Diabetes Obes Metab 2025).

A figure travels with its population and its timeframe or it does not travel at all. That cohort covers one drug at one point in time. It is not a class-wide or one-year figure, and it should not be stretched into one.

Two things follow.

The trial percentages describe doses many people never reach. Real-world escalation is slower than trial escalation, usually for good clinical reasons. If your own results were smaller than the trial numbers below, the likeliest explanation is not that you are a non-responder. It is that you were never at the dose those numbers describe.

A large share of the people who stop did not decide to. Nobody has measured that share cleanly, so it carries no percentage. What is well documented is that stopping is common and that the reasons are frequently not deliberate: cost, a rough patch after a dose increase that nobody helped anyone through, a pharmacy that did not have it, or simple drift.

If that is you, and you are watching the weight come back while trying to work out what you did wrong, the withdrawal evidence below is the long answer. Going back to your prescriber is an ordinary thing to do, not an admission.

How much weight comes back after stopping a GLP-1?

About 60 percent of the lost weight comes back within a year of stopping a GLP-1, according to a 2026 meta-regression, and regain then slows and levels off.

The withdrawal results are unflattering, and they get reported straight, because a softened version carries the greatest risk of hurting you. They are also more specific than "you gain it all back," and the specifics matter.

The best single quantitative estimate comes from a 2026 systematic review and nonlinear meta-regression pooling 48 studies, with six randomized trials and 3,236 participants feeding the model (Budini et al., eClinicalMedicine 2026). The Budini review describes regain as a curve rather than a number. Beyond the fast first year, the curve has two properties worth knowing separately.

• Regain decelerates. It does not run away. It slows on a characteristic half-life of roughly 23 weeks (95 percent confidence interval 17.3 to 34.3), so each successive stretch of time brings back less than the one before it.
• Regain plateaus below where you started. The model puts the plateau at about 75 percent of what was lost (95 percent CI 68.9 to 81.6), so roughly a quarter of the loss persists. For most trial participants, that worked out to something on the order of four to five percent of starting body weight, held long term.

The trajectories were broadly similar across liraglutide, semaglutide, and tirzepatide.

How reliable are the pooled regain figures?

Two cautions travel with those numbers. The plateau is modeled and extrapolated beyond the observed follow-up window, so it is a projection from a curve, not something anyone watched happen. It is also a population average, not a promise to any individual. The review was graded moderate rather than high in quality because only six randomized trials fed the quantitative model.

A separate meta-analysis of 18 randomized trials in 3,771 participants corroborates the magnitude from a different direction. It found a mean regain of about 5.6 kg in adults with obesity and confirmed that regain is time-dependent: it was substantially larger in trials that followed people longer (Tzang et al., eClinicalMedicine 2025).

Take the kilogram figure loosely. Heterogeneity in the Tzang analysis was near-total, meaning the underlying studies disagreed with each other enormously, so the pooled average is a rough central tendency rather than a precise number. Percentages are the more trustworthy unit for regain, and the figures that follow use them.

Two findings in the Tzang analysis are easy to misapply:

• Type 2 diabetes. In people with type 2 diabetes, much less weight came back, but blood sugar control deteriorated more.
• Semaglutide compared with liraglutide. The analysis reported more regain after semaglutide than after liraglutide. That is not evidence that one drug is worse to stop. The difference plausibly tracks how much weight was lost in the first place, and the paper does not settle the question.

Do you regain all the weight after stopping a GLP-1?

No. On average, people end up below their starting weight after stopping a GLP-1: STEP 1 extension participants were still a net 5.6 percent down two years in.

That is a smaller thing than the headline loss, and it is not nothing.

"You will regain everything" is both wrong and fatalistic, and the fatalism does its own damage. A reader who believes the loss was entirely temporary has less reason to stay in care, which is the one behavior with the best chance of preserving some of it.

The accurate version is more useful and no less frank. Regain is substantial, it is predictable, it slows, and a meaningful fraction of the benefit persists. Substantially attenuated is not the same as erased.

What did STEP 1, STEP 4 and SURMOUNT-4 find after people stopped?

STEP 1, STEP 4 and SURMOUNT-4 all found substantial regain after semaglutide or tirzepatide was withdrawn, in randomized trials run by two manufacturers on two different molecules.

That consistency is what makes the biological argument for regain airtight rather than merely sympathetic.

STEP 1 extension

Participants who completed sixty-eight weeks on semaglutide had lost a mean of 17.3 percent of their body weight. Both the drug and the structured lifestyle program were then withdrawn, and participants were followed for a further year. They regained 11.6 percentage points of that loss, ending at a net 5.6 percent below where they started: roughly two-thirds of the loss came back within a year, and the cardiometabolic improvements reverted toward baseline for most measures alongside the weight (Wilding et al., Diabetes Obes Metab 2022).

The treatment phase that came before is covered in semaglutide's evidence from three randomized trials.

STEP 4

After a 20-week run-in on semaglutide producing a mean 10.6 percent loss, participants were randomized either to continue the drug or to switch to placebo, with everyone keeping the lifestyle program. From that point to week sixty-eight, those who continued lost a further 7.9 percent, and those switched to placebo gained 6.9 percent. The difference between the two groups was 14.8 percentage points (Rubino et al., JAMA 2021).

SURMOUNT-4

After a thirty-six-week lead-in on tirzepatide producing a mean 20.9 percent loss, participants were randomized to continue or to switch to placebo, and followed for a further year. Those continuing lost another 5.5 percent; those switched to placebo gained 14.0 percent. At the end, 89.5 percent of the people who stayed on tirzepatide had held at least eighty percent of what they lost, against 16.6 percent of those who came off (Aronne et al., JAMA 2024).

Did people regain because they stopped dieting?

No. STEP 4 and SURMOUNT-4 both kept the lifestyle program running in the placebo arm, and both still saw substantial regain. Something pharmacological is happening, not only a lapse in behavior.

What the withdrawal trials cannot tell you

The STEP 1 extension withdrew everything at once. The lifestyle intervention stopped on the same day as semaglutide, so the extension is not a clean estimate of stopping the medication alone. It is arguably closer to what happens in the real world, and it is also a real limitation on the interpretation.

The same design explains an apparent discrepancy between the figures. STEP 1's roughly two-thirds regain sits a little above the pooled 60 percent, and it should: the STEP 1 extension pulled the lifestyle program at the same time as the drug, while the pooled estimate averages across trials that mostly did not. The two figures measure slightly different things, and the difference between them runs in the direction you would predict.

Trial participants did not choose to stop. Their study drug was withdrawn or swapped for a placebo. A person coming off deliberately, with a plan, with training established, and with continued follow-up is not in the identical situation.

Nobody has measured whether that changes the outcome, or by how much. Clinicians who work in this area believe it matters. That belief is reasonable, and it is not evidence.

Do the metabolic benefits of a GLP-1 last after stopping?

The metabolic benefits of a GLP-1 follow the weight rather than the drug after stopping, with one exception: blood pressure.

A post hoc analysis of SURMOUNT-4 looked at the people who stopped tirzepatide and sorted them by how much they regained. Participants who regained less than a quarter of their lost weight ended with waist circumference, non-HDL cholesterol, and fasting insulin that were not significantly different from their values at their lowest weight (Horn et al., JAMA Intern Med 2026).

The pattern is clean: hold most of the loss and you hold most of the benefit.

Why does blood pressure rise after stopping a GLP-1?

Systolic pressure rises about 7 mm Hg after stopping a GLP-1, even when most of the weight loss holds, because part of the blood pressure benefit came from the drug, not the weight loss.

Blood pressure is the exception to the weight-tracking pattern, and it is the most firmly established detail in the evidence on stopping. In the same SURMOUNT-4 post hoc analysis, systolic pressure rose by about 6.8 mm Hg even in the group that regained the least. Taken alone, that would be a post hoc analysis of a single trial with modest numbers in each stratum, worth no more than a raised eyebrow.

It does not stand alone. The 18-trial Tzang meta-analysis, an entirely different design working from a different set of studies, found that systolic pressure rose about 7.09 mm Hg after semaglutide was discontinued (Tzang et al., eClinicalMedicine 2025). Two independent methods landing within three tenths of a millimeter of mercury of each other is as close to corroboration as this literature gets.

What the convergence means is underreported, and it has a practical consequence. Part of the blood pressure benefit comes from the drug rather than from the weight, so it does not survive stopping, even for the person who does everything right and holds most of their loss. Your blood pressure can climb back while the scale barely moves.

That rise is not a failure of your weight maintenance, and trying harder at the weight will not fix it. It is a pharmacological effect ending, and it means a blood pressure medication dose reduced during treatment can quietly stop being correct.

Is stopping a GLP-1 linked to heart disease?

Stopping a GLP-1 within the first year was associated with higher coronary artery disease and heart failure risk in observational cohorts, according to a 2026 narrative review.

The regain numbers concern the scale and the markers that track it. This finding is a different order of consequence, and it deserves to be read as one rather than filed alongside them.

A 2026 review synthesizing the randomized withdrawal trials alongside real-world cohorts covering more than 289,000 patients reported that discontinuation within the first year of treatment was associated with increased risk of coronary artery disease and heart failure, compared with continuing. The authors frame discontinuation as "a high-risk clinical transition rather than a treatment endpoint" (Shah et al., Diabetes Obes Metab 2026).

Regaining weight is a setback. A heart disease signal is a different category of thing, and it is the strongest argument in this literature for why stopping belongs in a conversation with a prescriber rather than happening by not refilling a prescription.

How strong is the heart disease link?

The heart disease link is also the finding most easily overstated, so the calibration matters as much as the claim. The Shah paper is a narrative review rather than a systematic one, and the cardiovascular association comes from observational cohorts, not from randomized comparisons.

That design leaves a large opening for confounding by indication, the technical name for a simple problem. People who stop a medication within a year are systematically different from people who keep taking it, and the things that make someone stop are often the same things that predict a cardiac event. Cost, a new illness, other medications going wrong, a job loss, general disengagement from care: every one of those pushes both variables in the same direction, and an observational cohort cannot separate them.

So the finding does not support "stopping causes heart failure." It supports something narrower and still useful: the period around stopping is a window in which things go wrong, and it is the worst possible moment to be out of contact with a prescriber.

The unglamorous problem of other prescriptions that quietly stop being correct as your weight moves leads to the same conclusion from a completely different direction. Two weak-to-moderate arguments pointing the same way are worth more than either one alone.

Why does weight come back after stopping a GLP-1?

Weight comes back after stopping a GLP-1 because the drug opposed the weight your body defends rather than resetting it, and appetite rises while energy expenditure stays low.

The mechanism is not mysterious, and understanding it is the difference between reading regain as biology and reading it as personal failure.

A GLP-1 drug is a sustained load, not a replacement hormone

GLP-1 drugs do not correct a deficiency or restore your natural GLP-1. There is no hormone you were short of. A weekly GLP-1 receptor agonist holds a receptor system at concentrations your own physiology never produces, continuously, for the entire dosing interval.

Your own GLP-1 rises briefly after a meal and is destroyed within minutes. The drug is a sustained pharmacological load, which is a legitimate and effective way to treat something. It also means nothing durable is left behind when it goes.

Three changes that push toward regain

When the drug clears, three things move at once, and all three point the same way.

• Appetite comes back. The central signal suppressing it fades as the drug leaves.
• The appetite underneath is not your old appetite. It is the appetite of a body that has lost weight and is defending against that loss. In a study that followed people for a full year after diet-induced weight loss, ghrelin was elevated, leptin was suppressed, and several satiety hormones remained blunted, with subjective hunger significantly increased and the changes still present at one year (Sumithran et al., NEJM 2011). Those changes have nothing to do with any drug; they are what happens to a body that has lost a large amount of weight.
• Energy expenditure stays down. A smaller body costs less to run, and expenditure falls further than body composition alone predicts. That adaptation does not resolve on the same timescale as the drug clearing (Müller et al., Curr Obes Rep 2016).

The drug opposed a set point

A GLP-1 never resets a set point. It opposes one. While the drug is on board, it holds your weight below the level your biology defends. When it comes off, the biology it was opposing is still there, and it is now pushing harder than it was at baseline, because you are lighter: appetite up, expenditure down, drug gone.

That is why regain after stopping is fast and near-universal rather than slow and variable. It is also why the STEP 1 trialists concluded that their findings "confirm the chronicity of obesity and suggest ongoing treatment is required to maintain improvements in weight and health" (Wilding et al. 2022).

The comparison worth holding onto is not a course of antibiotics. It is blood pressure medication: nobody tells a person whose blood pressure normalized on an ACE inhibitor that they have graduated.

Is there a GLP-1 withdrawal syndrome?

No. GLP-1 drugs produce no withdrawal syndrome, because they suppress no hormonal axis that needs time to recover and cause no neuroadaptive withdrawal.

Most content about GLP-1 tapering invents something to justify it. Drugs get tapered when stopping them abruptly produces a specific, discrete harm:

• Corticosteroids are tapered because the adrenal axis has been suppressed and needs time to come back.
• Beta-blockers are tapered because abrupt withdrawal produces a rebound at receptors that have upregulated.
• Benzodiazepines, opioids, and antidepressants are tapered because they produce neuroadaptive withdrawal syndromes.

In every one of those cases, the taper exists to slow the rate of change, because the body cannot re-adapt fast enough. None of those conditions apply to GLP-1 drugs, and the reasons are specific, so the reassurance is real.

Why the testosterone analogy fails

A specific piece of misinformation circulates about GLP-1 withdrawal. People sometimes reason by analogy to testosterone, where taking an exogenous hormone shuts down your own production through a feedback loop, and coming off requires a recovery period. That analogy does not hold, and the reason is architectural.

Testosterone sits inside a closed feedback loop running through the hypothalamus and pituitary. Incretin hormones do not. GLP-1 is released from cells in your gut wall in response to nutrients arriving, with no hypothalamic regulator of that output and no established long-loop feedback from receptor occupancy back onto secretion (Santos-Hernández et al., J Mol Endocrinol 2024).

Nothing was suppressed. There is no recovery period to wait out, and no evidence of one.

What people experience after stopping is not withdrawal. It is the return of the untreated condition, arriving on exactly the schedule the pharmacology predicts. Feeling hungry again a few weeks after your last injection is not a symptom of coming off the drug; it is the drug finishing.

How long does semaglutide stay in the body after the last dose?

Semaglutide stays in the circulation for about five to seven weeks after the last dose, according to the Wegovy label, so a weekly GLP-1 injection already tapers itself.

The label's exact wording is that semaglutide "will be present in the circulation for about 5 to 7 weeks after the last" dose (Wegovy prescribing information, DailyMed, revised June 2026). Tirzepatide has a somewhat shorter half-life, so its washout is faster, though its label carries no equivalent sentence. Liraglutide, which is dosed daily rather than weekly precisely because its half-life is so much shorter, is essentially gone within days.

Why weekly injections leave slowly

The injectable weight-loss agents in this class are peptides engineered with a fatty acid chain that binds them tightly to albumin in your blood. That binding protects them from being taken apart and gives them a long half-life; semaglutide is more than 99 percent albumin-bound. The same piece of engineering is the reason a weekly injection works, the reason it takes weeks to reach full effect, and the reason it takes weeks to leave.

Albumin binding does not describe the whole class. The oral small-molecule agent approved in April 2026 is not a peptide, is not albumin-bound, and is taken daily, so none of the washout reasoning applies to it. Anyone on that daily tablet should read the washout reasoning as written about somebody else's drug.

Why stopping a weekly GLP-1 is not a cliff

Stopping a weekly agent produces a slow exponential decline that takes well over a month to complete, not a cliff. Halving a dose moves your blood concentration to roughly half; stopping outright passes through that same concentration on its own and then keeps going down. A step-down and a clean stop produce exposure curves that differ in their timing, not in their category.

A taper of a weekly GLP-1 is an attempt to make a gradual decline more gradual.

The absence of a cliff is a statement about the concentration of one drug in your blood and nothing else. It says nothing about the other prescriptions that stop being correct as the weight moves, which is a separate problem, managed by a prescriber.

Why do GLP-1 labels include restart rules but no taper instructions?

GLP-1 labels give detailed instructions for a gap in treatment, and several say when dose escalation has to be started over, but not one contains an instruction to taper on discontinuation.

That absence is the detail that settles the tapering question, and almost nobody notices it. It is not an oversight. The people who wrote those labels thought carefully about what happens when exposure to the drug is interrupted, because they wrote rules for it.

Each label sets a threshold beyond which a missed dose can no longer be made up, and after a longer gap the semaglutide and liraglutide labels direct that dose escalation be started over rather than resumed where it left off (Wegovy prescribing information, DailyMed, revised June 2026). The specific thresholds are printed on the labels and carried in the patient Medication Guides, so there is nothing to be gained by withholding them, and the companion guide to using GLP-1 medications well sets them out.

A missed dose is a question about continuing treatment. Stopping is a different and more consequential decision, so the guidance on stopping gives the reasoning, not the numbers.

What matters is the shape. The label writers considered the discontinuity from every angle, and they prescribed a ramp up with no ramp down.

Is any trial testing GLP-1 tapering against stopping?

Yes. The randomized REST trial is testing gradual semaglutide dose reduction against abrupt discontinuation, and primary completion is estimated for September 2028 (Yevusiak et al., PLoS One 2026; NCT07294950).

REST randomizes people who have lost weight on semaglutide, with percent body weight change as its primary outcome, and it is recruiting. Read the investigators' own language: they write that they "hypothesize" gradual reduction will be associated with less weight regain. That is a hypothesis, stated as one, by the people best positioned to know whether it has been established. It has not.

What has the discontinuation research tested so far?

Until REST reports, nobody has tested tapering against stopping. A 2026 systematic review and meta-regression of 48 studies of discontinuation found that exactly one included study used a dose taper, and in that study participants largely remained on some dose rather than stopping. The authors called for "individualised cessation strategies" because the evidence to specify one does not exist (Budini et al., eClinicalMedicine 2026).

A separate 2026 review states the position directly: structured multidisciplinary transition approaches may help "in the absence of validated tapering strategies" (Shah et al., Diabetes Obes Metab 2026).

Why step down slowly if no evidence shows tapering prevents regain?

A slow GLP-1 step-down serves three real purposes: appetite returns in stages, you stay in contact with a prescriber, and each step tests whether a lower dose is enough.

None of the tapering evidence means a step-down is pointless. It means the reasons for it are different from the assumed one, and the real reasons are better than the fake one.

Appetite comes back in stages rather than all at once. This rationale is behavioral and should be labeled as one, not a pharmacokinetic argument dressed up in physiology. Clinicians who do it describe it as considerably more manageable for the person going through it, and as giving them time to build eating structure while some pharmacological help is still present. That is plausible, and it is untested.

A step-down creates scheduled contact with a prescriber during the highest-risk window. The alternative, in practice, is that a person disappears from care on the day of their last injection. Regain builds over weeks and months, and it is far easier to respond to early. Nobody catches a drift that nobody is looking at.

The reason that matters most: each step shows whether a lower dose is enough. Clinicians who step people down rather than stopping them describe this as the single most valuable thing about going slowly, and several say they do it that way even when the plan is to stop entirely. Frequently, the answer that comes back is that a lower dose is sufficient. At that point the taper has stopped being a taper; it has become a maintenance dose.

The evidence on stopping, in two sentences

Every randomized withdrawal study, without exception, shows substantial regain and reversal of cardiometabolic benefit after stopping, and there is no trial evidence that any tapering method changes that trajectory. The step-down clinicians use exists to make appetite return manageable and to find the lowest effective maintenance dose, not to prevent regain.

That is not a rhetorical trick to talk you out of stopping. The last clause points at something with strong evidence behind it.

Does a lower tirzepatide maintenance dose hold the weight loss?

A lower tirzepatide maintenance dose holds substantially more weight loss than stopping and somewhat less than staying at full dose, according to the randomized SURMOUNT-MAINTAIN trial.

SURMOUNT-MAINTAIN is a 112-week randomized, double-blind, placebo-controlled trial that asked the question almost nobody asks. After an open-label weight-loss period on tirzepatide at each participant's maximum tolerated dose, 378 participants were randomized three ways: continued at that dose, assigned to a lower fixed dose and held there, or switched to placebo. Ninety-one percent completed (Horn et al., Lancet 2026).

At the end of the trial, measured from the original baseline:

• Continued at the maximum tolerated dose: 21.9 percent below baseline (95 percent CI 23.5 to 20.3)
• Assigned to the lower fixed dose: 16.6 percent below baseline (95 percent CI 18.0 to 15.1)
• Switched to placebo: 9.9 percent below baseline (95 percent CI 11.1 to 8.8)

All comparisons were P<0.0001. Participants were eligible for rescue treatment if they regained more than half of what they had lost. Eight percent of the maximum-dose group needed it, twenty-five percent of the reduced-dose group, and sixty-seven percent of the placebo group.

The middle row is the most practically useful finding in this whole body of evidence. A reduced maintenance dose is not a compromise between two real options. It is a real option, with randomized evidence behind it. The trial's own conclusion was that reducing the dose "might provide a valuable alternative to discontinuation, although individuals' treatment response might vary."

What are the limits of SURMOUNT-MAINTAIN?

SURMOUNT-MAINTAIN was a single trial, funded by the manufacturer of the drug it studied, in a population of a few hundred people, and the finding has not yet been independently replicated. The reduced-dose arm still needed rescue treatment three times as often as the group that stayed at full dose, so the trade-off is real rather than free.

Two practical considerations belong in that conversation, independent of the trial evidence. Adverse reactions in this class are dose-related, which the approved labels document, and cost scales with dose too. Whether either consideration outweighs the weight held at a higher dose is a judgment about one person's circumstances.

Is every dose on a GLP-1 label an approved maintenance dose?

No, and the point is counterintuitive: a dose printed on a label is not automatically an approved maintenance dose. At least one product in this class states outright that its lowest strength exists for starting treatment and is not approved for maintenance at all. A reduction landing there would describe an unapproved regimen using a number taken straight off the label.

Worse, the same dose can be approved maintenance for one indication and not for another, so two people on identical amounts of the identical drug can get different answers depending on why they were prescribed it. The destination is a decision made against your specific product and your specific reason for taking it. Nobody can reason it out from a dose chart, you included.

Is a lower maintenance dose the same as a GLP-1 taper?

No. Tapering means stepping down in order to stop, while maintenance means arriving at a lower dose and staying there, and only maintenance has randomized evidence behind it.

The taper-maintenance distinction is the single most conflated thing in this area. One plan has high-grade randomized evidence behind it, and the other has none. Consumer content routinely cites the maintenance data as though it validated tapering, and it does not, because SURMOUNT-MAINTAIN never tested a wind-down to zero.

SURMOUNT-MAINTAIN answers the question "can I take less?" It does not answer the question "how do I get off?"

That is the one operational point to take away. When someone offers you a taper protocol and cites trial evidence for it, the trial they are citing is almost certainly SURMOUNT-MAINTAIN, and SURMOUNT-MAINTAIN studied something else.

Which other medications need revisiting when you stop a GLP-1?

Insulin, sulfonylureas, and blood pressure medicines need revisiting when you stop a GLP-1, and water pills and narrow-margin drugs such as lithium deserve a look.

The other medications are where predictable harm clusters, and a gap in the system causes it, not anything GLP-1 drugs do.

Losing a significant amount of weight changes how much of your other medications you need. Blood pressure falls. Blood sugar falls. The body is smaller and eating less, so several medications that were exactly right a year ago become too strong, and a competent prescriber brings them down as the weight comes off.

Coming off runs that entire process in reverse, and the reverse gets managed far worse than the forward direction.

Insulin and sulfonylureas

Insulin and sulfonylurea doses reduced during weight loss will often need to go back up as the GLP-1 clears, over roughly the same weeks it takes to leave your system. A person who stops the GLP-1 and leaves the reduced diabetes doses alone is now undertreated. A person who adjusts them without supervision can go the other way into hypoglycemia, which is acute and can be severe. Neither direction is something to work out alone, and no article can give you a number for either.

Blood pressure medication

Blood pressure medication is the one that catches people who are otherwise doing everything right. As weight comes back on, blood pressure tends to follow it back up, which is the obvious half. The half nobody warns about: two independent analyses agree that systolic pressure rises by around 7 mm Hg after stopping, and the SURMOUNT-4 post hoc analysis found that rise even in people who held on to most of their weight loss. Part of that benefit was the drug, not the weight, so it leaves when the drug leaves.

Put that together with the deprescribing that happened on the way down and you get a specific, foreseeable problem. Someone whose blood pressure medication was reduced during treatment can have their pressure climb back afterward while their weight barely moves. Nothing about their weight maintenance has failed; their reduced antihypertensive dose has stopped being the right dose, and no symptom reliably announces it.

Blood pressure is also the cheapest and earliest thing anyone can watch, so ask your prescriber whether they want you tracking it at home through this period.

Water pills and narrow-margin drugs

Water pills were an accelerant on the dehydration risk while you were on the drug. Their dose was chosen for a body and an intake that are now changing, and it deserves a look in either direction.

Narrow-margin drugs such as thyroid replacement, blood thinners, seizure medications, and lithium do not all need changing, but they all deserve a look when body weight moves substantially.

Why do medication changes get missed after stopping a GLP-1?

Medication changes get missed after stopping a GLP-1 when different prescribers own different drugs, because then nobody owns the interaction.

If the weight medication came from one place and the blood pressure pills, the insulin, and the water pill come from somewhere else, the service that prescribed the weight medication does not manage your other prescriptions and often cannot see them. Your regular doctor does not necessarily know anything changed. Each party is doing its own job correctly, and the thing that needed doing falls between them.

That gap deserves to be described without blame. Telehealth prescribing has given a lot of people access to treatment that in-person care was not offering them, and plenty of those services are run well. But a service built around one medication structurally cannot deprescribe or re-prescribe the other seven, because they are not its medications. A service built around starting a medication has no particular reason to still be in your life on the day you stop.

What to tell the prescriber who manages your other medications

Tell the prescriber who handles the rest of your medications that you are coming off, in these words: "I am stopping a GLP-1 medication. Which of my other prescriptions should we be watching or adjusting as my weight changes?"

Then keep saying it, because your weight will keep moving for a year or more after your last dose. This is a running conversation rather than a single one. If nobody else is running it, you are the only person guaranteed to be in every one of these appointments.

What are legitimate reasons to stop a GLP-1?

Legitimate reasons to stop a GLP-1 include intolerable side effects, a serious adverse event, cost, supply interruption, pregnancy, a shift in risk and benefit, and reaching your goal.

Stopping is a legitimate decision, made with a prescriber. The regain evidence is not an argument that everyone should stay on these drugs forever; it is an argument that the decision should be informed. Each of the following is a real reason and deserves to be treated as a reason rather than as a failure.

• Side effects you cannot live with, after a real attempt at going slower and managing the symptoms. Staying longer at a dose is a labeled option, not a failure, and many people who quit do so during a rough patch that a slower pace would have solved. If that has been tried and the side effects are still intolerable, that is an answer.
• A serious adverse event. Pancreatitis, bowel obstruction, symptomatic gallbladder disease, or the very rare optic nerve injury that European regulators added to the product information in June 2025 (EMA safety committee, June 2025). These are stop-and-get-seen situations, not taper situations.
• Cost. Cost is probably the most common single reason in the United States, and it is not a moral failing. An account of stopping that omits cost is not describing the real world.
• Supply interruption. Supply interruption is also common, also not your fault, and it makes a planned wind-down moot for a large number of people. If you are off the drug because you were unable to get it, the framing about choosing to stop does not apply to you, but what to expect and which prescriptions to have looked at still do.
• Pregnancy, or planning one. GLP-1 medicines are not to be used in pregnancy, and the semaglutide labels set a stopping margin before conception.
• Risk-benefit has shifted. A new diagnosis, frailty, significant lean mass loss in an older adult, or an eating disorder emerging.
• You got where you wanted to go and want to try maintaining without the drug. That is legitimate as a choice. It should be an informed one, which is what the regain and maintenance evidence is for.
• You would rather not take an open-ended drug indefinitely. This reason deserves respect rather than an argument. Nobody has ten-year or twenty-year safety data on continuous use of these agents in people without diabetes, and the longest high-quality exposure dataset, the SELECT trial of semaglutide, ran a mean of about 40 months while measuring cardiovascular events rather than weight (Lincoff et al., NEJM 2023). A thoughtful person can look at that gap and decide against indefinite use, and that defensible position deserves to be weighed against the regain evidence rather than dismissed by it.

When should you stop a GLP-1 before trying to conceive?

The semaglutide labels direct stopping at least two months before a planned conception, a deliberate margin beyond drug clearance, because these medicines are not to be used in pregnancy.

A second reason gets almost no attention and matters just as much: a body in the middle of rapid weight loss is not in a good nutritional state for a first trimester. A reasonable pre-conception plan involves a stable weight and a normal intake for a while, not only a drug washout.

Fertility often improves as weight comes down, particularly for people with polycystic ovary syndrome. A person who has spent years assuming conception would be difficult can find that assumption no longer holds.

Do GLP-1 drugs reduce birth control pill absorption?

Only some do, because the interaction with oral contraception is specific to the molecule, not shared across the class. The tirzepatide labeling documents a substantial reduction in absorption of the hormones in a combined birth control pill, and directs users to switch to a non-oral method or add a barrier method after starting and after each dose increase (Zepbound prescribing information). Dedicated interaction studies of semaglutide found no clinically significant effect on either hormone (Kapitza et al., J Clin Pharmacol 2015; Jordy et al., Clin Pharmacokinet 2021).

Find out which molecule you are on, by active ingredient rather than brand name. Birth control is one of the few situations where the specific drug changes the answer.

Should you stop a GLP-1 before surgery?

Surgery usually calls for, at most, a temporary GLP-1 hold managed by the team doing the procedure, not a stop.

Tell every member of that team, every time, including for dental sedation and routine endoscopy. The guidance changed once already, in October 2024, moving away from telling everyone to hold the drug toward an individualized decision (multisociety guidance, 2024). That individualized approach only works if the individual details reach the person applying it.

Why was the suicidal thoughts warning removed from the Wegovy label?

The Wegovy label dropped its suicidal thoughts and behavior warning in February 2026, following an FDA request; an FDA review of 91 placebo-controlled trials found no increased risk.

The warning used to be there. The label's own list of recent changes shows the section marked removed, dated February 2026 (Wegovy prescribing information, DailyMed setid ee06186f, revised June 2026).

The removal followed a request from FDA the month before that also covered liraglutide and tirzepatide. FDA reported that its review of 91 placebo-controlled trials, covering 107,910 participants, found no increased risk of suicidal ideation or behavior, and no increase in other psychiatric adverse events either.

On the same label, in the same era, the boxed warning about thyroid C-cell tumors stayed exactly where it was, even though it rests on a rodent finding whose human relevance the label itself describes as undetermined.

What the label change means for the decision to stop

The label change bears directly on the judgment about long-term use. A safety system that demonstrably drops a warning when the evidence stops supporting it is one whose remaining warnings mean more, not less. It turns "this label has frightening warnings on it" into "this label has warnings that survived review," which is a different and more useful thing to know when you are deciding whether to keep taking something.

The change is also a live trap. A great deal of published material still describes the suicidality warning as current, and that is now a factual error. So do older copies of the label itself: an earlier version remains in circulation on a government website and still displays the warning as active. If a source you are weighing still carries it, that tells you how recently the source was checked, not something about the drug.

What happens when you restart a GLP-1 after stopping?

Restarting a GLP-1 after a full stop usually means climbing the dose escalation again from the bottom, side effects included, because gut tolerance fades without continuous exposure.

Almost nobody writes about restarting, and it is one of the most concrete costs of stopping entirely.

The tolerance your gut developed to these drugs was expensive to acquire. The nausea, the early fullness, and the reflux most people experience in the first weeks are largely driven by delayed gastric emptying, and that particular effect fades with continued exposure. The tirzepatide label documents it directly: the impact on gastric emptying was greatest after the first dose and had diminished by week six, at a higher dose (Zepbound prescribing information). The appetite effect, notably, does not fade the same way, which is the entire reason dose escalation works.

That adaptation was bought with continuous exposure. When exposure falls far enough for long enough, it is lost and has to be bought again, which is why several of the labels contain re-escalation rules after a gap and none of them contain a rule for winding down.

Does a short break mean starting the escalation over?

Not necessarily: a short interruption is a different thing. Each label sets a window inside which treatment is resumed as normal, and your prescriber or pharmacist can tell you in a minute which side of that window your gap falls on. Do not assume a short gap has cost you the ramp.

Has anyone tested cycling on and off a GLP-1?

No. No trial has examined stopping and restarting as a deliberate strategy, so whether repeated cycles carry a cumulative cost is open. Cycling is also the pattern real life produces most often, through cost and supply interruptions rather than through anybody's plan.

Why restarting is not a failure

The lost adaptation is a real cost, and it belongs in the decision. It is also why the choice between stopping and finding your lowest effective dose is not a semantic distinction: staying on a dose preserves the adaptation, and stopping surrenders it.

Restarting is not a failure. It is a normal part of managing a chronic, relapsing condition, and reinitiation is common in real-world data. The shame around restarting keeps people out of care for months while regain compounds, and that shame is entirely manufactured by a framing that was never true in the first place.

What predicts keeping weight off after stopping a GLP-1?

Staying on treatment is the strongest predictor of holding GLP-1 weight loss, by a wide margin over any habit or program studied so far.

The answer is uncomfortable in the other direction, so it comes first. In SURMOUNT-4, 89.5 percent of people who continued tirzepatide held at least eighty percent of their loss, against 16.6 percent of people who stopped (Aronne et al., JAMA 2024). Nothing else that has been studied comes close to that effect size. A list of habits presented as a substitute for the pharmacology is telling you something the trials do not support.

With that stated plainly, the other things are still worth doing, and one of them has real randomized evidence.

Does exercise help maintain weight loss?

Yes. In a four-arm randomized trial of weight-loss maintenance after an initial diet-induced loss, a supervised exercise program combined with liraglutide outperformed either intervention by itself. The combination reduced body fat percentage by 3.9 percentage points, roughly twice what exercise alone or the drug alone achieved, and only the combination improved HbA1c, insulin sensitivity, and cardiorespiratory fitness. The exercise-only group also maintained significantly better than placebo (Lundgren et al., NEJM 2021).

That is direct evidence that structured activity is protective during the maintenance phase.

Does resistance training preserve lean tissue?

Yes, during calorie restriction. A 2026 systematic review with network meta-analysis of thirty-four randomized trials found that adding exercise to calorie restriction preserved a mean 0.87 kg of fat-free mass and prevented roughly forty-six percent of the fat-free-mass loss, with resistance-containing programs outperforming endurance training (DOI 10.1111/dom.70873).

None of those trials were done in people taking GLP-1 drugs. Applying the finding to GLP-1 users is a reasonable inference from a mechanically identical situation, not a demonstrated result in this population. No completed randomized trial has tested resistance training in GLP-1 users, and one is enrolling. The evidence on preserving muscle during GLP-1 weight loss covers resistance training in more depth.

Which habits help people maintain weight loss?

Structural ones. Long-term maintainers in the National Weight Control Registry report a consistent cluster: regular activity, a fixed eating pattern that does not change between weekdays and weekends, eating breakfast, and self-monitoring weight (Wing and Phelan, Am J Clin Nutr 2005). The registry holds self-selected successes, so it describes what maintainers do rather than proving that doing those things causes maintenance.

Every habit on that list is structural rather than appetite-dependent: a fixed breakfast does not require wanting to eat less. That is the argument for building these habits while the drug is doing the hard part. When appetite is suppressed, adherence to a structure is nearly free, and when appetite returns, the structure is the only thing still standing.

What do clinicians see in people who hold more of their loss?

Clinicians describe four things that separate people who hold more of their loss from people who regain, and their account deserves the label of clinical observation rather than trial evidence:

• Training timing. Whether resistance training was established before stopping rather than planned for after.
• Eating structure. Whether eating structure was built while the drug was doing the work, rather than intake being low because nothing appealed.
• Pace of loss. How fast the weight came off.
• Staying in care. Whether the person stayed in care.

One more factor is the one to sit with: higher starting weight and longer duration of obesity mean stronger counter-regulatory drive. That is biology, not character.

Is the weight regained after stopping a GLP-1 mostly fat?

Nobody knows, because no published trial has measured the body composition of weight regained after stopping a GLP-1.

You will encounter a confident claim that the weight you lose is fat plus muscle while the weight you regain is preferentially fat, leaving you worse off after a cycle than before it. The claim is mechanistically coherent, and it is a reasonable hypothesis. It has never been measured in anyone who stopped one of these drugs.

The STEP 1 extension recorded body weight and cardiometabolic markers and ran no body composition assessment at all, and no published trial in this class has scanned anyone through a discontinuation. The source of the claim is an editorial proposing a mechanism, not a study reporting one. Treat it as unproven, and treat anyone stating it as established as having gotten ahead of the data.

Do GLP-1 drugs cause disproportionate lean mass loss?

No. Lean tissue made up about the same share of the weight lost on incretin drugs as on placebo or lifestyle intervention, so GLP-1 drugs are not uniquely catabolic.

The mirror image of the regain claim also needs correcting: the evidence on it is better, and it runs the other way. In the tirzepatide DXA substudy of SURMOUNT-1, roughly seventy-five percent of the weight lost was fat and twenty-five percent was lean tissue in the placebo arm as well as the drug arm (Look et al., Diabetes Obes Metab 2025). A meta-analysis of twenty randomized trials found the lean-mass fraction with incretin therapy not significantly different from lifestyle intervention alone (P=0.42) (Eisa and Barood, Diabetes Obes Metab 2026).

Lean mass loss is a property of losing weight, not a special toxicity of this class. Why GLP-1 muscle loss claims exceed the evidence looks at what lean-mass measurements include.

What should you ask your prescriber before stopping a GLP-1?

The most useful questions before stopping a GLP-1 cover your other prescriptions, a lower maintenance dose, warning signs, how often to check in, and restarting.

A generic plan is less useful than the set of questions that make the appointment work. Write these down:

• What happens to my other prescriptions when I come off, and who is going to manage that? Name insulin, any diabetes tablet, blood pressure medication, and any water pill specifically.
• Is there a lower dose that would hold most of this, and is that worth trying before we talk about stopping?
• What should I be watching for in the first couple of months, and what would you want to hear about?
• How often do you want to see me during this, and what are we measuring?
• What would make you want me to restart, and how would we handle that?
• If I do restart, do I go back to the beginning of the escalation?
• Is my blood pressure something we should be checking more often now, given that it tends to drift back up after stopping?
• Should we check my body composition before and after, so we know what changed?

If you have a history of disordered eating of any kind, that belongs in this conversation too, and it belongs with someone who knows that history. The return of appetite combined with fear of regain drives some people toward restriction, compensatory exercise, or purging. That is a reason to seek help promptly rather than a phase to ride out.

Why does food noise come back after stopping a GLP-1?

Food noise comes back after stopping a GLP-1 because the drug quieted a biological appetite drive that was there the whole time, and the drive returns when the drug leaves.

For most people, the hardest part of coming off is not the weight. It is the return of what patients call food noise: the constant background preoccupation with food that a great many people describe as having gone quiet, for the first time in their lives, while they were on the medication.

Its return is distressing. Clinicians report that patients cry about it in a way they never did about the number on the scale. Almost universally, people interpret it as proof of something about themselves: that they failed to keep it up, that they have no willpower, or that the drug was doing it, not them.

That last version is the most corrosive, because it retroactively invalidates a year of real work.

What does the return of food noise prove?

The self-blaming reading is exactly backwards. The quiet was real, and it was pharmacological: a drug that acts on the receptors regulating appetite reduced a biological drive, and the drive went down. When the drug left, the drive came back, because the drive was there the whole time.

Food noise was never a defect of character. It was physiology, and it was physiology before you ever took anything.

The fact that food noise disappeared on an appetite-regulating drug is the strongest evidence you will ever get that your struggle with food was biological rather than moral. People spend decades looking for that evidence and never find it. You ran the experiment on yourself, without meaning to, and it came back with an unambiguous answer.

What to do when food noise returns

Understanding the noise does not make it easier to live with, but it changes what the noise means and what to do about it. The answer is not to conclude that you are the problem and disappear from care. It is to go back to the person who prescribed for you, say that appetite has returned and the weight is moving, and ask what the options are.

Regain builds over weeks and months, and it is much easier to respond to early. A drift of a few pounds is a conversation. Two years of avoiding the scale out of embarrassment is a much harder problem, and shame is the mechanism that turns the first into the second.

The medication treated a chronic condition, and it worked for as long as it was there. That is not a disappointing result. It is how treatment for a chronic condition behaves, and it was never going to behave any other way.