Weight loss on a GLP-1 slows and then stops because the body pushes back, and for most people that happens within the first year and a half. In a large tirzepatide analysis, the median time to a plateau ran from 24.3 to 36.1 weeks depending on starting weight, and about nine in ten had plateaued by week 72 [1]. The plateau is the expected shape of treatment, and the trial evidence says the drug is still doing its job while the scale sits still, which is a different thing from what happens when it is stopped.
When the plateau arrives
The timing comes from a post hoc analysis of SURMOUNT-1 and SURMOUNT-4, the tirzepatide obesity trials, which defined a plateau as less than 5% weight change across a twelve-week window and every window after it [1]. It included only people who stayed on treatment and had already lost at least 5% by the trial’s main endpoint, 1,438 of them in SURMOUNT-1 and 259 in SURMOUNT-4, so it describes people for whom the drug was working.
Among them, the median plateau arrived at 24.3 weeks for people who started overweight, 26.0 for class I obesity, and 36.1 for classes II and III. By week 72, between 87.6% and 90.2% of each group had reached one. Higher doses of 10 or 15 mg, younger age and female sex all went with a later plateau, and SURMOUNT-4 showed the same pattern.
Diabetes trials show the same curve. A network meta-analysis of 55 placebo-controlled trials in 18,876 people with type 2 diabetes found the largest weight difference against placebo at 24 to 30 weeks, 2.42 kg, followed by a plateau period [4]. The effect on blood sugar held for at least 104 weeks, though by then the HbA1c reduction was about 0.36 points smaller than at its early peak, and the authors advise accounting for that weakening after two years.
Why the body stops the loss
Body weight behaves like a system with a thermostat: as weight falls, appetite rises, and intake creeps back toward what the new, lighter body needs. A published mathematical model of energy metabolism fitted that idea to the average weight curves from diet restriction, semaglutide 2.4 mg, tirzepatide 10 mg and gastric bypass [2].
In the model, every intervention except dieting substantially weakened the appetite feedback circuit, which is why the drugs and surgery produce a longer stretch of loss before the plateau. Gastric bypass also shifted the body’s equilibrium about twice as far as either drug. The author calls the results preliminary, and a model fitted to averages says nothing about a particular person’s curve, but it gives the plateau a mechanism that does not require the drug to wear off.
The drug is still working at the plateau
A direct test is a 60-week trial that randomized 120 adults 3:2 to semaglutide 2.4 mg or placebo [3]. At weeks 20, 40 and 60 the participants came into a laboratory and ate as much as they wanted from a provided lunch, with the calories measured. The semaglutide group ate 291.9, 240.2 and 269.5 kcal less than placebo at those three visits, so the gap at week 60 was about where it had been at week 20.
What changed was the feeling. Self-rated appetite, hunger and preoccupation with food all differed from placebo at week 20, and none of them did at weeks 40 or 60. Someone fourteen months in is probably comparing today’s hunger with their recent normal, which is already the drug’s normal. The sense that the medicine has stopped working can be sincere and still be wrong about what the drug is doing. How much less people eat in the first place is covered in the pooled intake evidence.
What people do at a plateau
Because higher doses went with a later plateau in SURMOUNT, the dose is the obvious question to raise with a prescriber. The the higher-dose evidence for semaglutide sets out what a step up buys and what it costs in side effects. The other thing worth watching through a long plateau is what the lost weight was made of, which is the subject of the muscle loss guide.
Stopping at a plateau carries the clearest risk, because the weight the drug is holding off tends to return, and it brings more than weight with it, as the stopping evidence shows. What none of these studies can yet say is how a plateau behaves beyond two years, or whether it drifts, because the trials that measured it did not run that long.