The two numbers, and why both are true
You have probably seen the frightening one: in the STEP-1 extension, people who stopped semaglutide regained about two-thirds of their lost weight within a year. SURMOUNT-4 found 82% of stoppers gave back a quarter or more of their loss. Those are real findings, and they describe people who stopped without any structured support.
Then there is the number nobody quotes: in Cleveland Clinic real-world data on roughly 8,000 discontinuers, about 45% maintained or kept losing at one year. Same drugs, same biology, different circumstances. Maintainers exist at scale. The gap between those two figures is not genetics. It is whether anything was in place when the medication ended.
Why regain happens: the math changed, not your character
The mask comes off
Your daily calorie burn, before and after
Illustrative figures for roughly a 50 lb loss. Your own numbers differ.
This is the part almost nobody explains properly, and understanding it removes most of the shame that makes regain worse.
- The drug never slowed your metabolism. Slowed stomach emptying is transit, not energy burn. The weight came off because appetite was muted and you ate less.
- The weight loss slowed your metabolism, as any method would. A smaller body burns fewer calories, and metabolic adaptation cuts burn a few hundred calories a day further as the body defends against perceived starvation. Lose 50 lbs and you may burn 300 to 500 fewer calories daily than your pre-loss self. Muscle lost along the way makes it worse, because muscle is the furnace.
- On the drug, that gap was invisible. Appetite suppression made it impossible to out-eat the reduced burn, so you never felt the deficit you were living in.
- Off the drug, the mask comes off all at once. Appetite rebounds to normal or above, aimed at a body burning hundreds fewer calories than the one that trained your instincts. Your old normal portion is now a surplus.
That is the whole mechanism. Regain is arithmetic arriving suddenly, not evidence about who you are. Which is good news, because arithmetic responds to structure, and character verdicts do not respond to anything.
The five levers, ranked by evidence
1. A protein floor (strongest, easiest)
Roughly 1.6g per kg of goal body weight daily, split into three meals of 30 to 40g. Protein is the most satiating macronutrient, so it directly blunts returning appetite, and it is the raw material that protects muscle. The per-meal split matters more after 45, because muscle becomes less responsive to smaller protein doses. One upside of stopping: for the first time in a year, your appetite will actually let you eat this much.
2. Resistance training, 2 to 3 short sessions a week
Not for calories burned. For the furnace. Without countermeasures, 25 to 40% of weight lost on a GLP-1 is lean mass, and regain preferentially returns as fat, which is how people end up with a worse body composition than they started with at a similar weight. In controlled research, adequate protein plus progressive resistance training cut lean-mass loss dramatically. Twenty to thirty minutes, full body, chair-supported variants for everything. Adherence beats intensity.
3. Recalculated targets
Off the drug, appetite intuition is gone and the old numbers are wrong. Estimate your new maintenance, then let a three-week trend correct the estimate. Numbers temporarily replace the signal the medication was providing, until a new normal forms.
4. A watched trend, not a watched day
Daily weigh-ins smoothed into a weekly trend line. Single days are noise (water, salt, hormones, sleep). The trend is signal. This is the lever that converts a slow drift into an early, small correction: caught at 4 lbs it is an adjustment, at 20 lbs it feels like starting over.
5. A plan for the food noise
It returns in weeks 2 to 6, often temporarily louder than before. Pre-loaded responses beat resolve, because resolve is exactly the resource this transition drains. The step-by-step version lives in how to stop food noise.
The first six weeks decide most of it
The transition is front-loaded. Drug levels fall for 3 to 5 weeks, appetite rebounds hardest in weeks 3 to 6, and habits formed in that stretch tend to persist. A rough map, and you can date it precisely with the washout calculator:
Expect the 2 to 6 lb water jump; change nothing. Set the protein target. Take starting photos and measurements.
Lock the protein floor. Start strength week one. Begin daily weigh-ins, watching the trend line only.
Appetite returns in earnest. Recalculate maintenance now, when the estimate starts to matter.
The hard zone. Tighten structure, not restriction. This is where most regain starts and where the plan has to be loudest.
The regain window in the trials. Adjust from real trend data. This is also when non-food improvements may drift back.
What else may come back (and nobody warns you)
If the medication quieted more than food, and for many people it did, some of that may return too. GLP-1 receptors sit in reward circuitry, so people commonly report reduced interest in alcohol, nicotine, and other compulsive pulls. There is essentially no research on post-discontinuation rebound of these non-food effects, but mechanistically it should happen. The same goes for physical improvements: joint pain, migraines, sleep, gut symptoms, cycle regularity.
Worth doing before you stop: write down everything that improved. Not just the weight. That list is what you are actually protecting, and it makes the case for structure far more vividly than a number does.
If you are already regaining
Nothing here is a verdict. Regain is the expected biological response to removing a signal, documented in every trial, and the community is full of people who caught it and turned it around. Three things help immediately: weigh in daily again even though you do not want to (avoidance is what turns 8 lbs into 30), restore the protein floor before anything else, and treat restarting medication as a legitimate option to discuss with your prescriber rather than as proof of failure. Restarting by decision at 5 lbs beats restarting by panic at 25.