Average weight loss on GLP‑1: two numbers from 12 trials

There is no single average. Each trial reports its own result, for its own drug, dose and length. In the STEP 1 trial (NEJM, 2021), adults on Wegovy (semaglutide 2.4 mg weekly) lost 14.9% of body weight at 68 weeks counting everyone who started, against 2.4% on placebo.3 The same paper estimates 16.9% if everyone had taken the drug as intended, against 2.4% on placebo.3

Table 1 lists nine trials of GLP-1 drugs approved in the US, with both averages where the trial reports them. Each figure is the mean change in body weight from the start of the trial, in percent, for the drug and for placebo. "Everyone who started" counts every person randomized, whether or not they kept taking the drug. "If everyone had stayed on the drug" is a model estimate. The last column names the source of each row, with its year. Read each row on its own: the trials enrolled different people for different lengths of time.

Table 1. Trials of approved GLP-1 drugs

Notes on three rows:

• SCALE. The paper reports weight lost with an older method, last observation carried forward: 8.4 kg on liraglutide against 2.8 kg on placebo at 56 weeks. It gives neither of the two averages in this table.11
• ATTAIN-1. A 2026 indirect comparison funded by Novo Nordisk gives -12.4% against -0.9% if everyone had stayed on the drug (the efficacy estimand). It also states that the 17.2 mg Foundayo tablet is equivalent to the 36 mg capsules of the trial.10
• OASIS 4. The "stayed on" figures come from the same 2026 indirect comparison funded by Novo Nordisk, not from the NEJM abstract.10

To turn a trial average into a weight timeline for one person, use the weight loss projection tool.

The two averages answer two questions. The first counts every person randomized, whether or not they kept taking the drug. The second estimates the result if everyone had taken the drug or placebo as intended. The STEP 1 trial of Wegovy (semaglutide 2.4 mg) reports both in the NEJM (2021): 14.9% and 16.9% weight loss at 68 weeks, against 2.4% on placebo in each analysis.3

The "stayed on" figure is not a simple average of the people who finished. In STEP 5, for example, it is an estimate for the whole randomized group: the paper describes it as the effect in all randomly assigned participants, "assuming that the drug or placebo was taken as intended".5

Each drugmaker names the two figures in its own way:

• Novo Nordisk, in its STEP UP release, uses the treatment policy estimand for everyone who started, and the trial product estimand for the "treatment effect if all people adhered to treatment".7 Its 2026 Wegovy HD release uses the names efficacy estimand and treatment-regimen estimand.8
• Eli Lilly uses the treatment-regimen estimand and the efficacy estimand, as in its TRIUMPH-1 release.15 On ClinicalTrials.gov, Lilly's efficacy analysis for SURMOUNT-1 excludes "data after prematurely stopping study drug".2 For ATTAIN-1 it uses data only "up to the earliest date of discontinuation of study drug or initiation of prohibited weight management treatments".14
• Boehringer Ingelheim defines its efficacy estimand as the effect "assuming patients remained on treatment for the entire study duration".16

So the two figures differ in how they treat the people who stopped the drug. Table 2 shows what the sources report on stopping, for each trial of Table 1. Each trial measured it in its own way, and each cell names the measure. The sources are the trial papers, the FDA labels and one indirect comparison funded by Novo Nordisk; the last column gives each one's year or label revision.

Table 2. How many people stopped the drug

In STEP 3, STEP 5, OASIS 4, SURMOUNT-1 and ATTAIN-1, fewer people stopped the drug than stopped the placebo, for any reason. In the same five trials, more people stopped the drug than the placebo because of adverse events (for OASIS 4, adverse reactions in the Wegovy label).4,5,17,19,1,20,10

Some drugmaker releases lead with the higher figure, the estimate if everyone had stayed on the drug. For retatrutide 12 mg in the TRIUMPH-1 trial, Lilly's press release of 2026-05-21 led with 28.3% weight loss at 80 weeks, against 2.2% on placebo.15 The NEJM paper, published online on 2026-09-29, reports 25.0% counting everyone who started, against 3.9% on placebo. Its abstract gives no "stayed on" figure.21

We found the 28.3% only in Lilly's release.15 We could read only the abstract of the NEJM paper. Lilly's release also gives the 25.0% figure, for the treatment-regimen estimand.15

Some releases give both figures. Novo Nordisk's STEP UP release of 2025-01-17 had this headline: semaglutide 7.2 mg "achieved 20.7% weight loss in the STEP UP obesity trial, and 18.7% regardless of treatment adherence". The 20.7% is the estimate if everyone had stayed on the drug, against 2.4% on placebo.7 Its OASIS 1 release of 2023-05-22, for oral semaglutide 50 mg, gave "15.1% weight loss (17.4% if all people adhered to treatment)".22 The 15.1% counts everyone who started, against 2.4% on placebo.23 When the US approved Wegovy HD, Novo Nordisk's release of 2026-03-19 led with "20.7% mean weight loss", the figure if everyone had stayed on the drug. A table in the same release also gives 18.7% for the treatment-regimen estimand, which counts everyone who started.8

The ATTAIN-1 and TRIUMPH-1 papers put the figure for everyone who started first. The ATTAIN-1 paper states its primary end point "according to the treatment-regimen estimand in the intention-to-treat population".13 The TRIUMPH-1 paper handles weight end points with "a treatment-regimen estimand (intention-to-treat [ITT])".21

Table 3 gives both averages for four more trials: of CagriSema, retatrutide and survodutide, and of oral semaglutide in the OASIS 1 trial. The last column names the source of each row, with its year. Each figure is the mean change in body weight in percent, drug against placebo. A figure marked "release" comes from the drugmaker's press release.

Table 3. Four more trials

Notes on Table 3:

• REDEFINE 1. Dosing was flexible. At week 68, 57.3% of people on CagriSema were on the highest dose, per Novo Nordisk.25
• TRIUMPH-1. Lilly's release calls retatrutide an investigational molecule, legally available only to people in Lilly's trials.15
• SYNCHRONIZE-1. The paper gives -12.2% for the 3.6 mg group and -13.0% for the 6.0 mg group, against -5.4% on placebo, counting everyone who started.26 Read the release's 16.6% only next to these two figures.16

Others have made this point before. Wharton and colleagues reviewed how stopping treatment and similar events change the results of weight-loss trials, in the International Journal of Obesity in 2021.27 Pharmacy Times gave both averages for TRIUMPH-2, another retatrutide trial, on 2026-09-30.28 Drug Topics reported on 2026-10-03 that tirzepatide's advantage over semaglutide 7.2 mg in an indirect comparison narrowed, and was not statistically significant, under the efficacy estimand.29 The GLP-3 Wiki's retatrutide tracker states that its phase 3 figures use the sponsor's efficacy estimand.30

How we built these tables. We checked 13 weight-loss trials in adults without type 2 diabetes: the nine in Table 1 and the four in Table 3. For each trial that reports it, we took the "everyone who started" figure from the journal paper. We took the "stayed on" figure from the paper when the part of it we could read gave one. If not, we took it from ClinicalTrials.gov results, a peer-reviewed secondary paper or the drugmaker's release, as each row's Sources cell shows. Of the 13 trials, 12 have both figures, and SCALE has neither. In all 12, people on the drug lost less when everyone who started is counted than in the "stayed on" figure. A second analyst downloaded the sources again on 2026-10-20 and compared 100 items: 97 matched, 3 stayed unconfirmed, and none were wrong. We read each figure on this page against its source again on 2026-10-25. The tables give no ranking: the trials differ in people, length, background care and comparator.

The trial results belong to the FDA-approved drugs. To compare programs that prescribe them, see the cheapest Zepbound programs, the cheapest Wegovy programs and the full ranking.

For how the next drug in this class compares, read retatrutide vs tirzepatide.

This article is for information only and is not medical advice. Talk to a licensed clinician before you start, change or stop a medicine.