The weight comes back gradually, and how much of it comes back varies enormously between one person and the next. The amount you keep off appears to decide how much of the health benefit you keep with it. SURMOUNT-4 is the trial that measured this, and it was built for the purpose rather than stumbling on it afterwards [1].
Everyone in the trial took tirzepatide for the first 36 weeks, openly and without a placebo group, which is how 783 people came to lose an average of 20.9% of their body weight before the real experiment began. At that point 670 of them were randomly assigned either to carry on with the drug or to switch to a placebo injection for another 52 weeks. Neither they nor their doctors knew which they had been given. That is a withdrawal design, set against the observational stopping this site examined in — a withdrawal design rather than the observational stopping this site examined in what happens in ordinary care.
What the scale did over the following year
The group that stayed on tirzepatide lost a further 5.5% on top of everything they had already lost, while the group switched to placebo gained 14.0% back. That leaves a gap of 19.4 percentage points between the two arms, with a confidence interval running from 21.2 to 17.7 [1].
Why two people who both stop can end up in very different places
A later analysis took the 308 people from the placebo arm who had lost at least 10% during the lead-in and sorted them by how much of that loss had returned by the end. That turns a single average into something far more useful for anyone actually deciding whether to stop [2]. The bands are shares of the weight each person had lost, so somebody in the largest band had regained three quarters or more of their own result rather than three quarters of their body weight.
| Share of lost weight regained | Waist gained | Systolic blood pressure |
|---|---|---|
| Under 25% | +0.8 cm (95% CI −1.0 to 2.6) | +6.8 mm Hg |
| 25% to under 50% | +5.4 cm | +7.3 mm Hg |
| 50% to under 75% | +10.1 cm | +9.6 mm Hg |
| 75% or more | +14.7 cm | +10.4 mm Hg |
The people who held on to most of their loss showed no measurable change in waist circumference at all, since 0.8 cm carries an interval running from minus 1.0 to 2.6. The people who regained the most put 14.7 cm back on. Non-HDL cholesterol behaved the same way, staying flat in the two smaller bands and rising 8.4% and 10.8% in the two larger ones.
The exception that is worth knowing about before you stop
Blood pressure did not follow that pattern. Systolic pressure rose in every single band, from 6.8 mm Hg among the people who barely regained anything to 10.4 mm Hg among those who regained the most. It came back even for the people who did best on the scale. Whatever the drug was doing to blood pressure, holding on to the weight loss did not preserve it.
One limit governs how hard any of this can be pushed, and it is worth being plain about: nobody was randomly assigned to a regain band. Everyone in that analysis received placebo, and which band a person landed in was an outcome rather than an allocation. So people who regained little may differ from people who regained a lot in ways that have nothing to do with the drug — the same problem this site set out in what happens when the control group never started treatment.
Is this specific to tirzepatide?
No, and the semaglutide evidence points the same way. When the STEP 1 extension followed people who had come off semaglutide, weight and the cardiometabolic markers both moved back toward where they had started. That is the pattern this site went through in the withdrawal evidence and again, across drug classes, in what comes back besides the weight [3].
What none of these trials can tell you is what happens to somebody who stops deliberately and carefully, with a plan, rather than because a prescription lapsed or a trial protocol switched them to saline. That trial has not been run, which is worth remembering the next time a page promises you a safe way off. — and the composition question underneath it, of what exactly is lost and what exactly returns, is in what is lost and what returns.