Coming off GLP-1 agonists — what to expect and how to taper

Most editorial coverage of semaglutide, tirzepatide, and liraglutide starts the conversation at "should I start" and ends at "see your doctor." The piece that's missing — and that millions of adults are now in — is what happens next. STEP 4 was the discontinuation extension of the semaglutide pivotal trial; people randomized off semaglutide regained roughly two-thirds of the lost weight within 48 weeks despite continued lifestyle counseling. SURMOUNT-4 replicated the pattern with tirzepatide. The rebound is the central feature of how these drugs work, not a side effect or a study artifact.

This playbook is for the person who has decided to come off, or who is being forced off by cost / supply / pregnancy / side effects, or who is planning a finite-duration use case (12–24 months) and wants to understand the trajectory in advance. It is not a recommendation to discontinue. For people with medical indications (BMI ≥30, sleep apnea on CPAP, T2D, established cardiovascular disease, NAFLD), continued therapy may be the right answer indefinitely.

The frame is harm reduction: the rebound is going to happen by default. The work below is what changes the default trajectory.

What "rebound" actually means — the trajectory

The STEP 4 and SURMOUNT-4 data tell a consistent story. After 20 weeks on semaglutide, participants randomized to placebo through week 68 regained on average ~6.9% of body weight while the continued-semaglutide group lost an additional ~7.9%. That ~14.8 percentage-point gap is the drug's continued effect. The placebo arm's regain rate was ~⅔ of the originally-lost weight over the 48-week off-period.

The trajectory inside that aggregate looks roughly like this:

• Weeks 1–4 off drug: Appetite returns. Hunger between meals, larger meal sizes, return of food noise (intrusive food-related thoughts). Weight stable or slowly rising.
• Weeks 4–12 off drug: Most of the rapid regain happens here. Set-point defense is the dominant biology — the body actively works to restore prior energy balance through increased appetite and (separately) decreased resting energy expenditure. Most people regain 30–50% of the lost weight in this window if no behavioral substrate is in place.
• Weeks 12–48 off drug: Slower regain trajectory. Either you've stabilized at a new (intermediate) set point because the behavioral substrate is holding, or you're trending back toward original baseline. The terminal state is highly variable across individuals.

The variables that determine where you land on this trajectory:

1. Behavioral substrate at the time of discontinuation. People who have used the on-drug period to establish stable eating patterns, training habits, and sleep discipline regain less than people whose drug-period was pure pharmacological appetite suppression.
2. Resistance training history and continuation. Lean mass loss during the on-drug period is largely irreversible without sustained training. Body composition shifts dramatically more for people who didn't train than for people who did.
3. Taper pace. Abrupt discontinuation produces sharper appetite return than gradual taper. The taper pace is not pharmacokinetically necessary (semaglutide has a 7-day half-life so it self-tapers over ~5 weeks regardless) but appears to affect the behavioral adjustment.
4. Concurrent life context. Discontinuation during stressful life events (job change, divorce, grief) produces worse trajectories. The drug had been compensating for emotional-eating patterns that re-emerge under stress.

The two legitimate use cases for discontinuation

Most discontinuation conversations fall into one of two patterns. The right preparation differs across them.

Pattern A: "I'm done — I've reached my goal weight and want off"

The honest framing the marketing skips: this is the harder use case. The data does not support this as a stable outcome for most adopters. People in this pattern who don't do the behavioral work during the on-drug period typically regain most of the loss within 12–24 months.

The preparation that changes the outcome:

• Resistance training 3+ sessions/week, established as a habit during the on-drug period. Not started after stopping — established by then.
• Protein 1.2–1.6 g/kg target body weight, 4 doses ≥30 g, established habit.
• Stable eating patterns — not "what I eat when not hungry" but a structured pattern (timing, meal composition, portion calibration) that has been practiced for 8+ weeks before discontinuation.
• Sleep adequacy (7+ hours consistent schedule).
• Stress baseline — discontinuing during high-stress periods predicts worse outcomes.
• Realistic regain expectation — most people stabilize 4–7 percentage points above their on-drug low. Anyone expecting to stay exactly at the on-drug nadir is going to interpret normal stabilization as failure.

Pattern B: "I'm being forced off — cost, supply, pregnancy, side effects"

Different preparation set. The timeline is often shorter — discontinuation isn't optional or timed.

• Pregnancy: Discontinuation is medically necessary. GLP-1 agonists are contraindicated in pregnancy (limited human data; preclinical concerns). The right pathway is clinical management with an obstetrician. The post-pregnancy resumption question is a separate conversation.
• Cost / supply: The most common forced-discontinuation pattern. Bridging options exist — liraglutide is cheaper than semaglutide; compounded versions exist (legality and quality both variable). The discontinuation strategy is the same as Pattern A but with less preparation time.
• Side effects: Severe GI, gallbladder events, pancreatitis, suspected thyroid-related concerns. These are clinical conversations. Discontinuation may need to be more abrupt than the structured taper below.

The structured taper (when discontinuation is planned)

The mainstream clinical advice is variable; the typical protocol involves stepping down through prior doses rather than stopping at the maintenance dose. For semaglutide (typical maintenance 2.4 mg / 1.7 mg / 1 mg / 0.5 mg / 0.25 mg starting doses), a structured taper looks like:

• Step 1 (4–8 weeks): Reduce from maintenance dose to next-lower dose (e.g., 2.4 mg → 1.7 mg). Maintain other variables.
• Step 2 (4–8 weeks): Reduce to next dose (e.g., 1.7 mg → 1 mg). Watch for appetite return.
• Step 3 (4–8 weeks): 1 mg → 0.5 mg. Most people notice meaningful appetite return at this step.
• Step 4 (4 weeks): 0.5 mg → 0.25 mg. Final pre-discontinuation dose.
• Step 5: Stop.

The total taper window is roughly 16–32 weeks. There is no rigorous trial evidence that this taper produces better outcomes than abrupt discontinuation — the rebound trajectory in STEP 4 and SURMOUNT-4 looks broadly similar regardless of taper. The case for the taper is behavioral: gradual reintroduction of appetite gives people more time to adjust eating patterns and to notice which behaviors are drug-sustained vs habit-sustained.

The case against the long taper: it extends drug exposure (cost, ongoing side-effect risk, ongoing IGF-1/glucose drift considerations) for benefit that isn't RCT-proven. Some clinicians advocate stopping abruptly and using the savings on a structured nutrition / training program for 12+ weeks post-discontinuation.

Both approaches are defensible.

Week-by-week: what to expect off drug

Generalized timeline for someone discontinuing from maintenance dose after 6+ months on therapy:

Weeks 1–2: Appetite begins to return but is still meaningfully suppressed (drug half-life ~7 days; full clearance ~5 weeks). Most people don't notice dramatic changes yet. This is the window to lock in habits.

Weeks 3–4: First clear appetite return. Hunger between meals. Food noise re-emerges in users who had it before therapy. GI symptoms (nausea, slow gastric emptying) resolve fully. Energy levels may improve (some people felt fatigued on therapy).

Weeks 4–8: Peak rebound risk window. Appetite often overshoots — the body is defending its set-point aggressively. Many people gain 3–6 lb in this window even with continued effort. This is normal physiology, not failure. The behavioral discipline matters most here.

Weeks 8–16: Stabilization or continued regain depending on substrate. If training + protein + sleep are dialed in, regain typically slows. If they aren't, regain continues.

Weeks 16–48: Long-tail trajectory. Either stable at a new (intermediate) baseline, or trending toward pre-therapy weight. The terminal state is roughly predicted by behavioral substrate quality during the on-drug period and the post-drug 16 weeks.

Weeks 48+: Maintenance phase. Most people stabilize 3–7 percentage points above their on-drug nadir if behavior is solid; significantly more if not. A meaningful minority maintain most of the loss; a meaningful minority return to or exceed baseline.

Biomarkers: what comes back when

Useful to track if you have access to bloodwork:

• Fasting glucose / HbA1c: Improvements on therapy may partially reverse with weight regain. HbA1c moves slowly (3-month average); fasting glucose moves within weeks.
• Lipids: Improvements partially reverse with regain. ApoB and triglycerides are the most sensitive markers; LDL is moderate; HDL is slowest to change.
• Blood pressure: Improvements on therapy partially reverse with regain. Track if hypertensive baseline.
• IGF-1 / GH axis: Not directly affected by GLP-1 therapy; this is more relevant if running concurrent peptide protocols.
• Liver enzymes (ALT, AST): NAFLD improvements on therapy may partially reverse. Worth tracking if NAFLD was the reason for therapy.
• Body composition (DEXA / InBody): The post-therapy DEXA is often the most informative data point — shows how much regain is fat vs lean mass. If lean mass is preserved, the regain is more cosmetic than metabolic.

The pattern in agressively-rebounding individuals: weight regain is heavily-fat-mass; lean mass remains depleted from the on-drug period; the post-regain body composition is worse than pre-therapy baseline despite similar scale weight. This is the worst-case trajectory and the reason resistance training during the on-drug period is non-negotiable.

The harm-reduction frame for discontinuation

The honest reality across the GLP-1 user population:

• Indefinite therapy is the FDA-approved framing. Wegovy, Zepbound, and the others are approved for chronic weight management — not for 6-month cycles.
• Most users will rebound substantially on discontinuation. This is the trial data.
• A meaningful minority will hold most of the loss. This minority typically did the behavioral work during the on-drug period and continued it after.
• The lean-mass loss is the irreversible part. Body composition at one year post-discontinuation is often worse than at therapy start, even when weight is similar.
• Cycling on and off ("12 weeks on, 12 weeks off") is not a well-studied protocol and the rebound timing makes it likely to produce repeated lean-mass loss without sustained weight benefit.

What the marketing skips: this is a chronic-therapy class of drugs that the cultural conversation has reframed as a temporary intervention. The trial data does not support that reframe.

What can be done about it: the rebound is mitigable but not eliminable. The interventions that mitigate are exactly the interventions that would have produced 5–10% weight loss without the drug — sustained caloric awareness, resistance training, protein adequacy, sleep, sustainable activity. The drug's value proposition is partly that it accelerates initial loss, partly that the on-drug period creates a window to establish the substrate. Whether the substrate gets established during that window is the load-bearing question.

What I'd do in your shoes

If you're considering starting with a planned finite-duration use case: enter the therapy with explicit behavioral milestones — training habit established by month 3, eating-pattern stable by month 6, sleep dialed in throughout. Treat the on-drug period as the substrate-building period, not just the weight-loss period. Plan the taper around month 9–12 if those milestones are met; extend if not.

If you're already on and considering off: honest substrate audit. Are you training? Eating predictably? Sleeping adequately? Can you predict your weekly food intake within 200 kcal? If yes, structured taper is reasonable. If no, the discontinuation will almost certainly produce substantial regain — and the right move may be either continuing therapy or pausing the discontinuation decision until the substrate is more solid.

If you're being forced off (cost, side effects, pregnancy): the timeline isn't optional, but the substrate work is the same. Prioritize resistance training, protein, sleep. Expect 4–8 weeks of significant appetite return. Track body composition (not just weight) post-discontinuation. Reassess at 12 weeks whether the trajectory is what you want.

If you've already discontinued and are regaining: this is normal physiology. The decision is not "did I fail" — it's "do I want to restart, or do I want to commit to the lifestyle floor at this weight and accept a new baseline." Both are defensible answers. Restarting after a regain period typically reproduces most of the original effect on the same dose.

What I would actively discourage: discontinuation without a behavioral substrate plan; cycling protocols (the rebound timing isn't compatible with short cycles); interpreting normal rebound as personal failure; assuming the drug "stopped working" when the actual story is that the body defends its set-point on discontinuation.

Operational notes for the taper

Not medical advice. Framed for someone who has decided to taper and is coordinating with their prescriber.

• Schedule the taper. Map the steps to a calendar. The structure helps maintain habits.
• Establish or re-establish the substrate during the taper, not after. Resistance training 3x/wk should be a habit by step 3, not started at week 1 post-discontinuation.
• Track weekly: body weight (same time, same conditions), waist circumference, training metrics (top sets on key compounds), subjective appetite 1–10, food noise 1–10.
• Get bloodwork at baseline, at the end of the taper, and 12 weeks post-discontinuation. Track ApoB, fasting glucose, HbA1c, lipid panel, ALT, total + free testosterone for men, fasting insulin.
• Body composition at start and end if accessible. DEXA is gold standard; InBody / BIA is trend-useful.
• Don't make the discontinuation a single big decision. Make it 4 smaller decisions (the taper steps), with re-evaluation built in.
• If the rebound trajectory becomes unmanageable in the post-drug period: restarting is a real option. There is no shame trajectory associated with chronic therapy for a chronic condition.

If you log these data points in the members area, the Evidence Engine aggregates across consented members. Over time the aggregate of GLP-1 discontinuation cycles with biomarker pairings becomes citable data — a community-version of the post-trial trajectory work that the pharmaceutical companies have less incentive to publish.

Sources and further reading

• Fat-loss decision guide — the upstream conversation if weight loss is the current goal
• Muscle preservation decision guide — the lean-mass conversation during therapy
• Semaglutide peptide page — full mechanism, dosing, safety, contraindications
• Tirzepatide peptide page — same structure for the dual-agonist
• GLP-1 receptor pharmacology dossier — the mechanism in depth
• Semaglutide muscle-loss critic response — the muscle-loss conversation in critic-response format

The STEP 4 (Rubino 2021), SURMOUNT-4, STEP 1 extension (Wilding 2022), and SELECT (Lincoff 2023) trials are findable on PubMed. The set-point defense literature underlying the rebound biology (Sumithran 2011 NEJM, Rosenbaum 2010 J Clin Invest, and the Hall 2016 Biggest Loser follow-up) is the deeper read on why discontinuation produces the trajectory it does.

— Last reviewed 2026-05-11.