Every GLP-1 program eventually runs into the same question: what happens if you stop? It is usually asked for one of three reasons — the cost stopped making sense, the supply ran out, or the goal weight arrived and staying on a weekly injection forever was never the plan. The honest answer is that this has been tested directly, in trials designed for exactly this question, and the results are unusually clear. They are also routinely overstated. Here is what the withdrawal trials found, and where the much larger real-world record disagrees with them.
The trials were built to answer this exact question
Most obesity trials ask whether a drug works. Three asked what happens when it is taken away, which is a different and more useful question — it is the design that isolates whether the weight loss belongs to the drug or to the person.
In the STEP 1 off-treatment extension, a subset of participants who had completed 68 weeks of once-weekly semaglutide 2.4 mg were followed for a further year after everything — drug and lifestyle program — was withdrawn. Mean weight loss at week 68 was 17.3%. By week 120, a year off treatment, they had regained 11.6 percentage points, leaving a net loss of 5.6% from where they started. The placebo group, which had lost 2.0%, ended at 0.1%.[1]
17.3%
Mean weight loss at week 68 on semaglutide 2.4 mg
STEP 1 extension
11.6 pp
Regained during the year after withdrawal
STEP 1 extension
5.6%
Net loss still held at week 120, a year off treatment
STEP 1 extension
STEP 4 ran the comparison inside the trial rather than after it. Everyone took semaglutide for a 20-week run-in, losing 10.6% on average; the 803 who reached the 2.4 mg maintenance dose were then randomized to continue or switch to placebo for 48 weeks. Those who continued lost a further 7.9%. Those switched to placebo gained 6.9% — a 14.8 percentage-point gap between staying and stopping. Waist circumference, systolic blood pressure and physical-functioning scores all improved with continued treatment relative to placebo.[2]
SURMOUNT-4 did the same with tirzepatide, and produced the starkest numbers in the set. After a 36-week open-label lead-in that averaged 20.9% weight reduction, 670 participants were randomized to continue or switch to placebo for 52 weeks. The continued group lost another 5.5%. The placebo group gained 14.0%. Put differently: 89.5% of those who stayed on tirzepatide still held at least 80% of their lead-in weight loss at week 88, against 16.6% of those who stopped.[3]
| Withdrawal trial | Weight change after stopping | Weight change if continued | Follow-up |
|---|---|---|---|
| STEP 1 extension (semaglutide 2.4 mg) | +11.6 pp regained | Not applicable — all participants stopped | 52 weeks off treatment |
| STEP 4 (semaglutide 2.4 mg) | +6.9% | −7.9% | 48 weeks after randomization |
| SURMOUNT-4 (tirzepatide 10 or 15 mg) | +14.0% | −5.5% | 52 weeks after randomization |
The cardiometabolic gains fade too
The number people fixate on is body weight, but the withdrawal data carry a second finding that matters more for health. In the STEP 1 extension, the improvements in cardiometabolic risk factors seen over the 68 treatment weeks reverted toward baseline by week 120 for most measures.[1] STEP 4 shows the mirror image while treatment continues: waist circumference fell a further 9.7 cm and systolic blood pressure a further 3.9 mm Hg versus placebo.[2] The benefit tracks the treatment. It is not a milestone you reach and then bank.
Why the real world looks so much flatter
Read only the trials and you would expect near-total regain to be universal. The largest look at what actually happens says otherwise. A retrospective cohort of 7,938 adults in a US health system, all of whom started injectable semaglutide or tirzepatide and then discontinued within three to twelve months, found a mean weight change from the point of discontinuation to one year later of just +0.5% among those treated for obesity — with, in the authors’ words, considerable individual-level variability.[4]
That is not a contradiction of the trials; it is a different population. Three things explain the gap. First, the depth of loss: this cohort had lost 8.4% on average by the time they stopped, against the 17–21% in the trial lead-ins — there was simply less to give back. Second, they did not vanish from care: within the year, 19.6% restarted the same medication and 35.2% received some other obesity treatment. Third, trial withdrawal is abrupt and total by design, which is not how most people stop.
What follows in practice
The first implication is framing: these medications behave like treatment for a chronic condition, not a course of antibiotics. Stopping does not undo the loss overnight, but it removes the mechanism that was holding it. Anyone planning to stop is better served planning what replaces it.
The second is that the lowest dose that holds is a real conversation to have with a prescriber, and a different one from stopping — see what the trials show about GLP-1 maintenance dosing and, for the smaller-dose approach some clinics market, the evidence on microdosing GLP-1s.
The third concerns body composition. A meaningful share of weight lost on a GLP-1 is lean mass, and weight regained is disproportionately fat, so a stop-start cycle is not composition-neutral — the detail is in GLP-1s and lean-mass loss, and the countermeasure, resistance training with adequate protein, is the same one used against age-related muscle loss.
Finally, if cost is the reason for stopping, it is worth exhausting the alternatives first: coverage, appeals and the compounded market are covered in GLP-1 cost and insurance coverage, and the current prices across telehealth providers sit on our GLP-1 provider comparison.
The honest bottom line
Stop a GLP-1 and you should expect to regain a substantial share of what you lost — that is the consistent finding across three randomized withdrawal trials, and the cardiometabolic improvements go with it. But “you will gain it all back” overstates it: the trials withdrew people abruptly from losses of 17–21%, and the real-world average over the first year off treatment is much closer to flat. The useful question is not whether stopping causes regain. It is what you put in its place.