There is no evidence-based answer to this question, and the useful thing an article can do is say so precisely. No randomised trial has ever compared one duration of GLP-1 treatment against another. Every published trial either runs to a fixed endpoint — 68 weeks, 72 weeks, 104 weeks, 208 weeks — or randomises people to continue versus stop; a search across ClinicalTrials.gov and PubMed in August 2026 found no exception. What does exist is the longest exposure data in the field and a well-described shape to the weight curve, and both are routinely misread. This article covers what has actually been measured, what the plateau does and does not mean, and where the widely quoted two-year figure comes from.
Weight plateau: the point at which weight loss slows to near-flat. The only formal definition used in this literature is a change of less than 5% over a 12-week interval, and over every 12-week interval after it.
Has anyone tested how long to stay on it?
No. Two trial designs dominate this field, and neither one asks the question.
The first is fixed-duration. STEP 1 ran to week 68, SURMOUNT-1 to week 72, STEP 5 to week 104, SELECT to week 208. Everyone in the treatment arm receives the same length of treatment, so the trial can say what happened over that period and nothing about whether a different period would have been better.
The second is randomised withdrawal. STEP 4 and SURMOUNT-4 put everyone on treatment for a set lead-in and then randomly assign them either to continue or to switch to placebo. That design answers "is continuing better than stopping" — decisively, as it turns out — but it compares treatment against no treatment, not one duration against another.
Nobody has randomised people to a shorter course versus a longer one and compared the results. The searches behind that statement were run six ways on ClinicalTrials.gov and separately on PubMed in August 2026, and every interventional study returned was one of the two designs above. So a confident, specific answer to "how long should I be on this" is not coming from the trial literature, because the trial that would produce one has not been done.
How long have people actually been treated?
About four years is the longest continuous exposure published, and it comes from a trial that was not a weight-loss trial.
SELECT randomised 17,604 adults and its prespecified weight analysis reports that weight loss "continued over 65 weeks and was sustained for up to 4 years", reaching a mean reduction of 10.2% at week 208 against 1.5% on placebo, alongside a 7.7 cm reduction in waist circumference (Ryan 2024).
The population matters more than the number. Everyone enrolled was aged 45 or over, had a BMI of 27 or above, and had established cardiovascular disease — a prior heart attack, a prior stroke, or symptomatic peripheral artery disease. Type 1 and type 2 diabetes were both exclusions. Mean age was 61.6 years, 72.3% were men, and mean baseline BMI was 33.3 (Ryan 2024). This is a secondary-prevention cardiovascular trial rather than a weight-management one, and its 10.2% is materially smaller than what the weight-management trials report. It should not be read as what a weight-management patient would expect.
One further precision. "Four years" is the scheduled endpoint, not everyone's exposure: mean follow-up was 39.8 months and mean treatment exposure 34.2 months (Lincoff 2023). We go through what that trial was designed to measure in long-term effects of GLP-1.
The longest run in a weight-management population is half that. STEP 5 followed 304 adults with obesity, or overweight with a weight-related comorbidity, and without diabetes, for two years: mean weight change of −15.2% at week 104 against −2.6% on placebo (Garvey 2022). It is a small trial and 93% of participants were white, so it is not a Singapore-representative population.
When does the weight stop falling?
Later than most people assume, and the answer is drug-specific.
For semaglutide, the description comes from individual trial narratives rather than a formal analysis. STEP 5's authors write that weight showed a "substantial initial reduction… which plateaued after approximately week 60 and was maintained for the remainder of the study", and the trial's own anchor points bear that out: −15.6% at week 52 and −15.2% at week 104 (Garvey 2022). SELECT's full text describes the trajectory as continuing to week 65 and then being sustained through week 208 (Ryan 2024). Both are descriptive wordings from the authors — "approximately", "appeared to" — not a statistical time-to-plateau, and the two trials do not use the same definition.
For tirzepatide there is a dedicated analysis. A post-hoc study of SURMOUNT-1 and SURMOUNT-4 defined plateau as a weight change of less than 5% over a 12-week interval and over all subsequent 12-week intervals, and found median times to plateau of 24.3, 26.0, 36.1 and 36.1 weeks across the overweight, class I, class II and class III BMI categories. By week 72, between 87.6% and 90.2% of participants had plateaued — later, on average, for those on higher doses, younger participants and women (Horn 2025). Three caveats travel with those numbers: the analysis includes only participants who were adherent and who reached at least 5% weight loss, so it describes responders; seven of the ten authors are employees and shareholders of the manufacturer and the other three report consulting or advisory income from it; and there is no equivalent published analysis for semaglutide, so the tirzepatide medians and the semaglutide narratives are not interchangeable.
In Singapore tirzepatide's current registrations carry both a type 2 diabetes and a weight-management indication, while the 2023 injection registration SIN16718P carries type 2 diabetes only (Singapore NDF). Everything above is reported as a trial result rather than as an outcome to expect.
Does a plateau mean it is time to stop?
No, and this is the most consequential misreading in the whole topic.
A plateau marks the point where weight stops falling. It does not mark the point where the medicine stops acting. In STEP 5, weight flattened around week 60 and the reduced weight was then held for another full year while treatment continued (Garvey 2022). In SELECT, the curve flattened at week 65 and the reduction was held for three more years on treatment (Ryan 2024). In both cases the flat line is being produced by continuing treatment.
What happens when treatment is removed is measured separately, and it runs the other way. In STEP 4, participants switched to placebo gained 6.9% of body weight over 48 weeks while those continuing lost a further 7.9% (Rubino 2021). In SURMOUNT-4, the placebo group regained 14.0% over the following year against a further 5.5% lost on continued treatment (Aronne 2024). In the STEP 1 extension, participants who came off treatment regained roughly two-thirds of their loss within a year, with most cardiometabolic improvements reverting towards baseline (Wilding 2022).
A flat curve on treatment and a flat curve off treatment are two different things. Why two-thirds of the weight comes back goes through the regain data in full.
Where does the two-year figure come from?
England's NHS, and it is a funding decision rather than a clinical one.
NICE Technology Appraisal TA875 recommends semaglutide for weight management "only if: it is used for a maximum of 2 years, and within a specialist weight management service", alongside BMI criteria and a suggestion to "consider stopping semaglutide if less than 5% of the initial weight has been lost after 6 months of treatment" (NICE 2023).
NICE is a health-technology assessment body, not a medicines regulator. The cap reflects the trial evidence available at the time and cost-effectiveness modelling for NHS funding; the appraisal does not claim that two years is the medically correct length of treatment. It carries no legal or clinical standing in Singapore.
The two NICE appraisals also disagree with each other. TA1026, covering tirzepatide, contains no maximum-duration sentence anywhere in its recommendations, using instead a review at six months on the highest tolerated dose if less than 5% of initial weight has been lost (NICE 2024). The appraisals were published nearly two years apart under different commercial arrangements. A figure that changes between two appraisals of the same drug class is a funding parameter, not a finding about the medicine.
Does diabetes change the numbers?
Substantially — which is why trial figures should never be lined up beside each other without stating who was in each one.
Semaglutide 2.4 mg at 68 weeks produced a mean weight change of −14.9% in adults without diabetes (Wilding 2021) and −9.6% in adults with type 2 diabetes (Davies 2021). Tirzepatide 15 mg at 72 weeks produced −20.9% in adults without diabetes (Jastreboff 2022) and −14.7% in adults with type 2 diabetes (Garvey 2023). The gap is large enough that a figure quoted without its population is close to meaningless.
Every duration and plateau dataset in this article — SELECT, STEP 5, the SURMOUNT plateau analysis — comes from a population without diabetes. None of them describes what someone with type 2 diabetes should expect on any of these questions. GetLean is a weight-management service; anyone with diabetes has that condition managed by their own doctor, and the duration question belongs in that conversation.
So what actually decides duration?
In practice, not a rule — and the real-world data makes that plain.
Among 125,474 US adults starting one of these medicines, 64.8% of those without type 2 diabetes had stopped within a year, against 46.5% of those with type 2 diabetes. Larger weight loss on treatment predicted lower discontinuation; moderate or severe gastrointestinal side effects predicted higher (Rodriguez 2025). In a separate cohort of 7,881 adults with overweight or obesity and without diabetes, 80.8% ended up on a lower-than-maximal maintenance dose, and mean weight reduction at one year was 8.7% — against 13.7% for semaglutide and 18.0% for tirzepatide among those who stayed on treatment at a high maintenance dose (Gasoyan 2025).
Both are observational electronic-health-record studies in a US insurance and access context that does not transfer to Singapore, and neither can separate cause from circumstance: someone on a lower dose may be there because of side effects, cost, access or a clinician's judgement. What they do describe is that duration in practice is settled by tolerability, cost, access and clinical judgement — because there is no duration rule available to settle it.
At GetLean, treatment duration is reviewed at the monthly check-in with Dr Quek. That check-in is the standing prompt for the conversation, and a patient can raise it at any check-in rather than wait to be asked; whether to continue, adjust or stop is his clinical decision at that review.
At GetLean, our philosophy is that GLP-1 medication should act as a catalyst — not something to depend on indefinitely. That is our position, and it is a response to this gap rather than a reading of the trials: nothing in the literature specifies a length of treatment, so the length gets decided by the goal, the response and the doctor's judgement, and the work of making an exit survivable happens long before the question is asked.
Individual results vary, and clinical-trial figures describe the populations studied. How long to stay on a prescribed medication is a decision to make with the doctor who prescribed it.
Common questions
How long should you stay on GLP-1 medication?
There is no evidence-based answer. No randomised trial has ever compared one duration of treatment against another — every published trial runs to a fixed endpoint or randomises people to continue versus stop. Duration is therefore a clinical decision made with the prescribing doctor, based on the goal, the response and the whole medical picture.
When does weight loss plateau on GLP-1 medication?
Later than most people expect, and the figures differ by drug. STEP 5's authors describe semaglutide weight loss as plateauing after approximately week 60 (Garvey 2022), and SELECT's weight analysis describes loss continuing to about week 65 (Ryan 2024). The only formal analysis is for tirzepatide, where median time to plateau ran from about 24 to 36 weeks depending on starting BMI class (Horn 2025). Individual results vary.
Does hitting a plateau mean the medication has stopped working?
No. The plateau marks the point where weight stops falling, not the point where the medicine stops acting. In STEP 5 the reduced weight was then held for another year on treatment (Garvey 2022), and in SELECT for three more years (Ryan 2024). What the withdrawal trials show is that the lower weight is held while treatment continues.
Is there a two-year limit on GLP-1 medication?
Not as a clinical rule. The two-year figure comes from England's NHS funding guidance, which recommends semaglutide for weight management only if it is used for a maximum of two years within a specialist service (NICE TA875). That is a health-technology funding decision rather than a safety or efficacy finding, and it has no standing in Singapore. The later tirzepatide appraisal sets no maximum duration at all (NICE TA1026).
What is the longest anyone has been treated in a trial?
About four years, in SELECT — where weight loss continued to about week 65 and was then sustained to a mean −10.2% at week 208 (Ryan 2024). That trial enrolled adults with established cardiovascular disease and overweight or obesity without diabetes, so it is not a weight-management population and its figures should not be read as one.