GLP-1 Trial Duration: What 68, 72 and 176 Weeks Can Tell You

Most GLP-1 weight-loss approvals rest on trials lasting about 68 to 72 weeks, which is roughly a year and a half. Longer studies now exist, including a two-year semaglutide trial, a three-year tirzepatide analysis in people with prediabetes, and a heart outcomes trial that followed people for over three years on average, and each timeline answers a different kind of question.

If you are planning to take a GLP-1 for years, that gap between "a year and a half of data" and "the rest of your life" is worth understanding. Not to scare you; just so you know which questions are well answered and which are still being worked on.

How long were the main GLP-1 trials?

Trial	Medicine	How long	The question it answered
SURMOUNT-OSA	Tirzepatide	52 weeks	Does it reduce sleep apnea severity?
STEP 1	Semaglutide 2.4 mg	68 weeks	How much weight do adults without diabetes lose versus placebo? (-14.9% vs -2.4%)
SURMOUNT-1	Tirzepatide	72 weeks	Same question for tirzepatide (-15.0% to -20.9% by dose, vs -3.1%)
STEP 5	Semaglutide 2.4 mg	104 weeks	Does the weight loss hold over two years? (-15.2% vs -2.6%)
SELECT	Semaglutide 2.4 mg	Mean follow-up 39.8 months	Does it reduce heart attacks, strokes and cardiovascular deaths?
SURMOUNT-1, prediabetes group	Tirzepatide	176 weeks plus 17 weeks off	Does it hold weight loss for three years and delay type 2 diabetes?

These durations are not random. Each was chosen to fit its endpoint (the outcome a trial is designed to judge; see GLP-1 trial endpoints explained). Weight takes months to plateau, so weight trials run past a year. Heart attacks and strokes are rarer, so SELECT needed 17,604 people and years of follow-up to count enough of them.

What can a 68- or 72-week trial tell you?

Quite a lot, honestly:

• How much weight people lose on average, and how many reach 5%, 10% or 15%. In STEP 1, 86.4% of the semaglutide group lost at least 5% by week 68.
• The common side effects and when they happen. Nausea and diarrhea were the most common with semaglutide and were typically mild to moderate and temporary. SURMOUNT reports describe gastrointestinal effects happening mostly during dose escalation.
• Whether the curve flattens. Both trials show weight loss slowing to a plateau, which is useful context if you are in one now (see GLP-1 weight loss plateau).

What can a 68-week trial not tell you?

• Whether the effect lasts for many years. It tells you about week 68, not year six.
• What happens after stopping. Unless the design includes an off-treatment period, the trial ends while everyone is still on the drug.
• Rare harms. If a side effect happens to 1 in 10,000 people per year, a trial of 2,000 people for 18 months may never see it. Those signals usually come from bigger, longer studies and post-marketing surveillance.
• Outcomes that take years to develop, like heart attacks, kidney failure or cancer.

A good example of the last point: GLP-1 labels carry a boxed warning that these medicines cause thyroid C-cell tumors in rodents, and that it is unknown whether they do so in humans. That "unknown" is partly a time-and-numbers problem: a rare cancer needs very large, very long observation to rule in or out. Our thyroid warning explainer covers what that warning does and does not mean.

What did the longer trials add?

Two years (STEP 5). Weight loss with semaglutide was -15.2% at week 104, close to the 68-week result in STEP 1. That supports durability while on treatment, at least to two years. Note the size, though: 304 participants, which means the result is less precise than a 2,000-person trial.

Three years (SURMOUNT-1, prediabetes group). Of SURMOUNT-1's 2,539 participants, 1,032 also had prediabetes, and they stayed on their assigned dose for 176 weeks. Weight change was -12.3%, -18.7% and -19.7% on the three tirzepatide doses versus -1.3% on placebo. Type 2 diabetes was diagnosed in 1.3% of the tirzepatide groups versus 13.3% on placebo. The analysis reported no new safety signals.

Then the clever bit: a 17-week off-treatment period. After stopping, 2.4% of the tirzepatide groups had type 2 diabetes versus 13.7% on placebo. So part of the benefit persisted briefly after stopping, but the number of diagnoses in the tirzepatide groups also edged up once treatment ended. A short trial could never have shown that; a trial that ended at week 176 without the off period could not either. (More in GLP-1 medicines and prediabetes.)

Three-plus years of events (SELECT). Heart outcome trials are long by nature. SELECT followed people for a mean of 39.8 months, and its result (fewer major cardiovascular events) is the kind of thing you simply cannot learn in 68 weeks.

Does a longer trial always mean better evidence?

Not automatically. Long trials have their own problems:

• People stop treatment. In SELECT, 31% of the Wegovy group and 27% of the placebo group permanently stopped the study drug, according to the Wegovy label. The longer a trial runs, the more the "drug" group becomes a mix of people on and off the drug.
• Life happens. Over years, participants change other medicines, diets and doctors. That is realistic but noisier.
• Fewer people may be followed. The three-year SURMOUNT-1 result applies to the prediabetes subgroup, not the whole trial.

So duration is one dial, not a quality score. A big 72-week trial and a small two-year trial each bring something.

How to read "long-term" claims in the news

When an article says "long-term results," ask:

1. How long, exactly? Weeks or years, and was that a fixed length or an average follow-up?
2. On or off the drug? Results "during treatment" and "after stopping" are different questions. (Our guide to extension studies covers the "after" part.)
3. Everyone, or a subgroup? The three-year tirzepatide data came from 1,032 people with prediabetes.
4. What kind of outcome? Weight, a lab value, or real clinical events?

And keep the week number glued to every figure you repeat. "Lost 20%" is a different statement from "lost 20% at week 72."

Where your own timeline fits in

Trials end. Your treatment may not. Keeping a long, boring, honest record (weights, doses, side effects, labs) is how you and your clinician build your personal "long-term data." If you log those in GLPily, the trends stay in one place across months and years instead of scattered across appointment notes.

Common questions

How long were the Wegovy and Zepbound weight-loss trials?

The main weight trials were 68 weeks for semaglutide 2.4 mg (STEP 1) and 72 weeks for tirzepatide (SURMOUNT-1). Longer trials and follow-ups now exist, up to 176 weeks for tirzepatide in people with prediabetes.

Is there long-term safety data on GLP-1 medicines?

There are multi-year trials and years of post-marketing experience, especially for diabetes use. Rare or very slow-developing effects remain under study, which is why labels and FDA communications keep being updated.

Why don't companies just run 10-year trials?

Keeping thousands of people on placebo for a decade raises ethical and practical problems, and many participants would stop or switch treatment. Long-term questions are usually answered by combining long outcome trials with real-world data.