The head-to-head finally exists. It ran 72 weeks, enrolled about 750 adults with obesity and no diabetes, and produced the single number that has defined this argument ever since: 20.2 percent mean body-weight reduction on tirzepatide versus 13.7 percent on semaglutide. That is SURMOUNT-5, published in 2025. It is a clean trial, and it answers the least important question on the table.
Here is my thesis. Tirzepatide wins on scale, and it is not close. Semaglutide owns the evidence that a cardiologist actually reads. The gap that should move a formulary committee is not six and a half points of body weight over 72 weeks. It is 17,604 patients followed for nearly 40 months in a placebo-controlled trial with a hard cardiovascular endpoint, which Novo Nordisk has and Eli Lilly, for this class of question, does not.
I think the market has priced the weight number and ignored the evidence asymmetry. Every consumer explainer of the past year leads with 20.2 versus 13.7. Almost none of them mention that the two molecules are not competing on the same evidentiary terrain at all, and that on the terrain that determines who lives and who has a stroke, only one of them has run the race.
Some taxonomy first, because the brand names in the headline are doing damage. Tirzepatide is sold as Mounjaro for type 2 diabetes (FDA approved May 2022) and as Zepbound for obesity (November 2023). Semaglutide is sold as Ozempic for diabetes (2017) and as Wegovy at 2.4 mg for obesity (June 2021). Same molecule, two boxes, two labels, two prices.
The trial everyone cites did not compare the drugs everyone names
SURMOUNT-5 was not Mounjaro versus Wegovy. It was Zepbound versus Wegovy: the obesity dose of tirzepatide against the obesity dose of semaglutide, in people without diabetes. Mounjaro was not in the room.
The trial that did put Mounjaro against semaglutide is SURPASS-2, from 2021, and it is a different animal. It compared tirzepatide against semaglutide at 1 mg: the diabetes dose, not the 2.4 mg obesity dose. Tirzepatide at 15 mg cut HbA1c by 2.30 percent against 1.86 percent for semaglutide, and dropped 11.2 kg against 5.7 kg, over 40 weeks.
Those are large margins. They are also, quietly, a comparison against a comparator running at roughly 40 percent of its eventual obesity dose. When a partisan cites SURPASS-2 as proof that tirzepatide dominates semaglutide, they are citing a trial that never tested the dose semaglutide is famous for. This is the most common error in the entire genre, and it appears in outlets that should know better.
Six and a half points is a real number, and it is a surrogate
Body weight is an intermediate endpoint. It is measured because it is cheap, fast, and correlated with the things we actually care about: infarction, stroke, kidney failure, death. Correlation is not the endpoint. Torcetrapib raised HDL beautifully and increased mortality. The history of cardiovascular medicine is a graveyard of surrogates that moved in the right direction while patients did not.
I want to be precise, because this is where the argument usually collapses into cheap skepticism. The 20.2 percent figure is not fake, and weight reduction at that magnitude is very likely to translate into outcome benefit. The point is narrower: likely is a different epistemic object than demonstrated in 17,604 people against placebo, and we should not let the first borrow the authority of the second.
Compare the anchor trials on their own terms. SURMOUNT-1 put tirzepatide at 20.9 percent weight reduction over 72 weeks in 2022. STEP-1 put semaglutide at 14.9 percent over 68 weeks in 2021. SURMOUNT-5 then confirmed the ordering head-to-head. Three trials, one consistent finding, no serious dispute: on the scale, tirzepatide is the stronger molecule.
SELECT is the asset, and Lilly does not have one
In 2023, Novo published SELECT: 17,604 participants with established cardiovascular disease and overweight or obesity, no diabetes, randomized to semaglutide 2.4 mg or placebo, followed a median of nearly 40 months. The result was a 20 percent reduction in major adverse cardiovascular events, a hazard ratio of 0.80. In March 2024 the FDA added cardiovascular risk reduction to the Wegovy label.
That label is the most underrated asset in this entire market. It converts an obesity drug into a cardiovascular drug, which changes who prescribes it, which formularies cover it, and which specialty society writes it into a guideline. It is also why some plans that dropped weight-loss coverage in 2025 kept covering semaglutide for the cardiovascular indication.
Lilly's cardiovascular outcomes program, SURPASS-CVOT, was built as a non-inferiority comparison against dulaglutide, an active GLP-1 comparator, not placebo. That is a defensible, even ethical design. It is also a design that can only ever answer not worse than another GLP-1. It cannot generate the sentence "reduces major adverse cardiovascular events by 20 percent versus placebo," because it never ran that experiment.
Add the kidney data and the asymmetry widens. The FLOW trial, in semaglutide at the 1 mg diabetes dose, cut major kidney outcomes by roughly a quarter in people with type 2 diabetes and chronic kidney disease. That is the semaglutide molecule accumulating organ-protection evidence across territories while the tirzepatide file, on those same territories, is thinner.
Where Lilly does have hard endpoints, it wins clean
This is not a story about one company doing homework and the other not. It is a story about two companies choosing different homework.
In December 2024 the FDA approved tirzepatide for moderate-to-severe obstructive sleep apnea in adults with obesity, the first drug ever approved for that indication. That is a hard clinical endpoint, prospectively tested, won outright. The SUMMIT program in heart failure with preserved ejection fraction produced its own real results in a population where almost nothing has ever worked.
So the honest read is not a ranking. It is two evidentiary portfolios with different shapes. Semaglutide is deep on atherosclerotic cardiovascular disease and kidney outcomes. Tirzepatide is deep on weight magnitude, sleep apnea, and HFpEF. Anyone who tells you one drug is simply "better" has compressed a portfolio comparison into a scalar, and the compression is where the information dies.
The strongest case against my own thesis
Let me argue the other side properly, because it is stronger than most of its proponents make it.
First: absence of a placebo-controlled cardiovascular outcomes trial is not evidence of absence of benefit. By the time SELECT reported, running a large placebo arm in high-risk cardiovascular patients had become ethically fraught. Lilly's non-inferiority design against dulaglutide may reflect equipoise, not evasion. Punishing a sponsor for the trial it could no longer ethically run is a strange way to read evidence.
Second: the surrogate objection cuts both ways. Weight reduction of 20 percent is not a modest signal on a soft endpoint. It is a large effect on a variable with a dense causal literature linking it to the outcomes in question. If you believe obesity causes cardiovascular disease, and the entire premise of SELECT is that treating it helps, then a molecule that treats it harder has a strong prior in its favor.
Third: class effects are real. If the cardiovascular benefit of semaglutide runs substantially through GLP-1 receptor agonism and weight reduction, both of which tirzepatide does at least as well, then the mechanistic expectation is that tirzepatide's outcome data, when it fully arrives, will not disappoint.
I find that third argument genuinely persuasive, and I still think it loses. Mechanistic expectation is what we had for niacin, for fibrates, for hormone replacement in the 1990s. The class-effect prior is exactly the reasoning that outcomes trials exist to discipline. It is usually right. It is spectacularly, expensively wrong often enough that we built an entire regulatory apparatus around not trusting it.
What an honest read looks like
Tirzepatide is the more powerful weight-reduction molecule, demonstrated head-to-head, and that is settled. Semaglutide is the molecule with a placebo-controlled hard-outcomes trial in 17,604 people and a cardiovascular indication on its label, and that is also settled.
These two sentences are both true, and almost every version of this comparison you will read drops one of them. The 20.2-versus-13.7 framing is not wrong. It is a single frame from a longer film, replayed until people mistook it for the plot.
My prediction, stated so it can be checked against me later: within three years the interesting question will not be which molecule loses more weight. It will be whether the outcomes evidence for tirzepatide arrives in the shape the class-effect argument predicts, and what happens to Wegovy's cardiovascular franchise when it does. That is the trial to watch. The scale was never the story.