Tirzepatide vs semaglutide: what SURPASS-2 measured

The trial, stated plainly

What it was: an open-label, 40-week, phase 3 trial registered as NCT03987919 and published in the New England Journal of Medicine in 2021. Assignment was double-blind with respect to which tirzepatide dose a patient received.

Who was in it: 1879 adults with type 2 diabetes and inadequate glycemic control on stable metformin alone. At baseline the mean glycated hemoglobin was 8.28%, mean age 56.6 years and mean weight 93.7kg, with a mean diabetes duration of 8.6 years and a mean BMI of 34.

What was compared: tirzepatide 5mg, 10mg and 15mg once weekly against semaglutide 1mg once weekly, added on top of metformin in every arm.

Why this trial and not a comparison table: both drugs were dosed inside one protocol, in one population, and measured at one timepoint. Numbers assembled from two separate trials do not have that property, and stitching them together is the failure mode this page exists to avoid.

What the primary endpoint returned

The endpoint: change in glycated hemoglobin from baseline to week 40. Every figure below is that endpoint, in that trial, at that timepoint.

What the authors concluded: tirzepatide at all three doses was noninferior and superior to semaglutide on the primary endpoint, with the 5mg difference at P = 0.02 and the 10mg and 15mg differences at P below 0.001.

The size of the gap: between the best tirzepatide arm and semaglutide, 0.45 of a percentage point. Real, statistically significant, and smaller than the difference between either drug and its own baseline.

What body weight did in the same 40 weeks

The absolute changes: the prescribing information reports week 40 body weight change in this trial at -5.7kg on semaglutide 1mg, and -7.6kg, -9.3kg and -11.2kg on tirzepatide 5mg, 10mg and 15mg.

The differences as published: least-squares mean estimated treatment differences against semaglutide 1mg of -1.9kg, -3.6kg and -5.5kg, all at P below 0.001. The signs are negative in the source, and they rederive from the arm figures: -7.6 minus -5.7, -9.3 minus -5.7, and -11.2 minus -5.7.

The context that keeps this honest: weight was a secondary measure in a glycemic trial, in a population selected for type 2 diabetes on metformin. It is a real result from a real head-to-head, and it is not an obesity trial result.

The other measure in the same table: fasting serum glucose fell 49 mg/dL on semaglutide and 55, 59 and 60 mg/dL across the tirzepatide arms.

What the safety reporting showed

The dominant category: gastrointestinal, primarily mild to moderate in both drugs. Nausea ran 17% to 22% across the tirzepatide arms against 18% on semaglutide.

The rest of the common events: diarrhea at 13% to 16% against 12%, and vomiting at 6% to 10% against 8%.

Hypoglycemia: blood glucose below 54 mg/dL was reported in 0.6% of the 5mg group, 0.2% of the 10mg group and 1.7% of the 15mg group, against 0.4% on semaglutide.

Serious adverse events: 5% to 7% across the tirzepatide arms and 3% on semaglutide. That gap and the efficacy gap come from the same trial and the same population.

Which one lowered HbA1c more in SURPASS-2?

Tirzepatide, at every dose tested. The estimated mean reductions were 2.01, 2.24 and 2.30 percentage points at 5mg, 10mg and 15mg, against 1.86 percentage points for semaglutide 1mg, and the authors reported all three as noninferior and superior on the primary endpoint.

The margin in plain terms: between 0.15 and 0.45 of a percentage point, with confidence intervals that stayed below zero in all three comparisons.

The qualifier that travels with it: against semaglutide dosed at 1mg. That comparator is a fact of this protocol, and every figure on this page is bounded by it.

Which doses did the trial actually compare?

Three tirzepatide arms at 5mg, 10mg and 15mg once weekly, against a single semaglutide arm at 1mg once weekly. Roughly 470 patients sat in each arm of the modified intent-to-treat population.

Why the comparator matters so much: a single-dose comparator arm sets a ceiling on what the trial can say. SURPASS-2 measured tirzepatide against semaglutide 1mg and reports nothing about semaglutide at any other dose.

Where the doses themselves come from: the trial protocol, not from us. Our tirzepatide guide and semaglutide guide lay out each compound's own trial record separately.

Does SURPASS-2 answer the weight-loss question?

Partly, and only for this population. Every patient enrolled had type 2 diabetes and was already on metformin, with a mean BMI of 34 and a mean baseline weight of 93.7kg, so the weight numbers describe that group at week 40 and nobody else.

What it does report: a 5.7kg fall on semaglutide 1mg and an 11.2kg fall on tirzepatide 15mg, with treatment differences of -1.9kg, -3.6kg and -5.5kg as the source prints them.

What would be a mistake: setting those figures beside a number from an obesity trial of either drug. SURPASS-2 enrolled adults with type 2 diabetes on metformin and measured weight as a secondary endpoint over 40 weeks, so a figure from a trial with a different population or a different endpoint is not a like-for-like comparison.

What a single head-to-head cannot settle

Four things, all visible in the design. The trial was open-label, so patients and investigators knew which drug was given. It ran 40 weeks, so nothing longer is in evidence here.

The comparator ceiling: semaglutide appeared at one dose. Any statement about the two molecules at other doses is outside this trial's reach.

The population boundary: adults with type 2 diabetes on metformin, with a mean diabetes duration of 8.6 years. Results in that group are results in that group.

What we did with the rest: left it out. Every trial figure on this page comes from SURPASS-2 or from the label section that reports SURPASS-2, and no figure has been imported from another study to fill a gap.

What did SURPASS-2 report on side effects?

Gastrointestinal events dominated both drugs and were primarily mild to moderate. Nausea ran 17% to 22% across the tirzepatide arms against 18% on semaglutide, diarrhea 13% to 16% against 12%, and vomiting 6% to 10% against 8%.

The serious-event line: 5% to 7% on tirzepatide against 3% on semaglutide, reported in the same paper as the efficacy figures.

Hypoglycemia below 54 mg/dL: 0.6% at 5mg, 0.2% at 10mg and 1.7% at 15mg, against 0.4% on semaglutide 1mg.

What happened to fasting glucose in the same trial?

It fell in every arm, and the ordering matched the HbA1c result. Table 5 of the prescribing information reports a week 40 change of 49 mg/dL on semaglutide 1mg against 55, 59 and 60 mg/dL on tirzepatide 5mg, 10mg and 15mg.

The baselines behind those figures: 171 mg/dL on semaglutide and 174, 174 and 172 mg/dL across the three tirzepatide arms, so the groups started close together.

Why it belongs beside the primary endpoint: fasting glucose and HbA1c measure different windows of the same control problem, and both moved the same way in this trial.

More from the library

Around this breakdown: five pages covering the two compounds on their own, and the machinery both arms ran on.

• Tirzepatide dosage guide: the full trial record for the compound in the three-arm side of this comparison.
• Semaglutide dosage guide: the same treatment for the comparator arm.
• Subcutaneous vs intramuscular: the route both arms of this trial used, and what the labels say about it.
• Injection sites in the trials: the regions the protocols and label sections record.
• Micrograms vs milligrams: the unit trap that ruins comparisons before the pharmacology gets a chance to.

Once a week: one head-to-head read of this kind goes out in the weekly briefing, free.

Sources

What this page is built from: one trial report, its registry entry, and the label section that reports the same trial. Nothing else, by design.

Last reviewed: 27 July 2026.

1. 1

   Mounjaro (tirzepatide) injection, for subcutaneous use. Prescribing information, Eli Lilly and Company. Section 14.3, SURPASS-2, and Table 5. DailyMed SPL, set id d2d7da5d-ad07-4228-955f-cf7e355c8cc0.

2. 2

   Frias JP, Davies MJ, Rosenstock J, et al. Tirzepatide versus Semaglutide Once Weekly in Patients with Type 2 Diabetes. N Engl J Med. 2021;385(6):503-515. doi:10.1056/NEJMoa2107519. PMID 34170647.

3. 3

   A Study of Tirzepatide (LY3298176) Versus Semaglutide Once Weekly as Add-on Therapy to Metformin in Participants With Type 2 Diabetes (SURPASS-2). ClinicalTrials.gov identifier NCT03987919. clinicaltrials.gov/study/NCT03987919.