A semaglutide weight loss plateau is a period with little further change in weight after an earlier reduction. A stable lower weight can represent maintenance. The measurement alone does not show that every treatment effect has disappeared.
The timing is a group finding measured from the trial baseline, not a deadline for an individual. STEP 5 and the longer SELECT analysis help answer when average loss slowed, what was maintained and which people were studied.
When did weight loss plateau in STEP 5 and SELECT?
STEP 5 randomized 304 adults with obesity, or overweight with a weight-related condition, without diabetes. Both groups received a behavioral intervention. The study's discussion describes a plateau after approximately week 60; the main week-104 estimates were −15.2% with semaglutide and −2.6% with placebo.
The SELECT weight analysis followed a different population: 17,604 adults with established cardiovascular disease and overweight or obesity, without diabetes. Average loss continued to about week 65, then was sustained through week 208. The reported in-trial mean changes at week 208 were −10.2% and −1.5%, respectively.
Scroll horizontally to read all columns.
| Question | STEP 5 | SELECT weight analysis |
|---|---|---|
| Participants | 304 adults without diabetes. | 17,604 adults with established cardiovascular disease, without diabetes. |
| When average loss leveled off | Approximately week 60. | After continued loss to about week 65. |
| Reported follow-up endpoint | 104 weeks. | 208 weeks. |
| Semaglutide mean weight change | −15.2% at week 104. | −10.2% at week 208, in-trial analysis. |
| Placebo mean weight change | −2.6% at week 104. | −1.5% at week 208. |
These were trials of subcutaneous semaglutide 2.4 mg under defined protocols. The table is a comparison of study contexts, not a head-to-head test or a prediction for every semaglutide product, amount or patient. Population, accompanying care, follow-up and analysis differ.
Why can weight loss plateau while an effect persists?
Think of the difference between the level of weight and its rate of change. An illustrative index that falls from 100 to 85 and remains near 85 has stopped falling, but it remains 15 units below its original baseline. Describing only the flat interval hides the earlier reduction that was maintained.
A 2024 mathematical modeling study by Kevin Hall explored how energy intake and appetite feedback could produce different plateau timings. It fitted average trajectories and suggested that a plateau can arise despite a persistent modeled intervention effect. These are preliminary model-based results, not a way to diagnose why one person's weight has stopped changing.
The trial curve alone cannot identify an individual's explanation or prove a particular mechanism of “resistance.” A flatter average line is an observation that needs interpretation, not a laboratory test showing that a drug has become inactive.
Plateau, little initial response and weight regain are different
Scroll horizontally to read all columns.
| Pattern | What the measurements show | What to check in the evidence |
|---|---|---|
| Little initial response | Limited reduction from the starting baseline. | The population, observation period and analysis used to define response. |
| Plateau after a reduction | Little further change while an earlier reduction remains. | The starting reference and the duration of maintenance. |
| Regain | An increase from a previously lower weight. | Whether treatment continued or had been withdrawn, and which baseline was used. |
The distinction matters when comparing continued-treatment trials with weight regain after stopping semaglutide. A withdrawal study changes the treatment condition. It cannot be substituted for an explanation of a plateau during continued treatment.
Why a study average does not predict your own timeline
An unchanged group mean can combine people who are still losing weight, people maintaining and people gaining. It does not describe a single “typical participant” followed by everyone else. Short-term variation in one person's measurements is also different from a sustained average pattern over many trial visits.
Check who remains in the analysis. “Completed the trial” is not the same as “remained on assigned treatment throughout.” In STEP 5, the primary treatment-policy analysis included randomized participants regardless of treatment discontinuation or rescue intervention. An analysis restricted to a period on treatment addresses a different question.
SELECT likewise reported in-trial and first on-treatment analyses. Mixing a value from one analysis with a value from the other can create a misleading comparison. A useful summary names the analysis instead of selecting the largest weight-loss percentage.
Does a plateau mean treatment should change?
STEP 5 and SELECT do not provide an individual rule for increasing, switching or stopping a medicine. Their group curves cannot determine the cause of one person's plateau or weigh that person's benefit against adverse effects. Medication decisions require an assessment with the prescribing clinician; this article provides no adjustment schedule.
A long elimination half-life is not an explanation of a year-long weight trajectory either. The half-life vs duration of action guide separates drug exposure from the clinical outcomes measured over follow-up.
Which question should you read next?
If the question is what changed in body composition, read the lean mass and weight-loss evidence. For cardiovascular outcomes in SELECT's population, use the SELECT heart-outcomes guide. The FLOW kidney-outcomes guide concerns a separate population with type 2 diabetes and chronic kidney disease.
For a plateau claim, keep the baseline, time interval, treatment condition and analysis beside the graph. That lets you distinguish an initial response, a maintained lower weight and subsequent regain without turning a group average into a promise or a treatment instruction.
