Most GLP-1 comparisons are assembled from separate trials run years apart in different populations, and the honest ones say so. This one does not have to be. Novo Nordisk makes both semaglutide and liraglutide, and in STEP 8 it put them in the same trial against each other.
That makes this one of the few comparisons in the category where the answer rests on a single randomised protocol rather than on a reconstruction — and the answer is not close.
The trial
STEP 8 (NCT04074161) randomised 338 adults with overweight or obesity to semaglutide 2.4 mg once weekly, liraglutide 3.0 mg once daily, or placebo, over 68 weeks. It completed in March 2021 and its results are posted in the registry, which is where every figure below was read on 2026-09-11.
Retention was high and even across arms: 120 of 126 completed on semaglutide, 118 of 127 on liraglutide, 81 of 85 on placebo. Nobody died in any arm. The trial is small by the standards of the pivotal obesity programmes, but its design is the thing that matters — the comparison is internal.
The result
| Semaglutide 2.4 mg weekly | Liraglutide 3.0 mg daily | Placebo | |
|---|---|---|---|
| Mean weight change, week 68 | −16.4% | −6.4% | — |
| Reached ≥10% loss | 83 of 117 | 30 of 117 | 12 of 78 |
| Reached ≥15% loss | 65 of 117 | 14 of 117 | 5 of 78 |
| Injections per year | 52 | 365 | — |
Semaglutide produced 2.6 times the mean weight reduction.
The threshold figures matter more than the means, because averages conceal how differently the two drugs behave at the far end. At 10%, roughly 71% of the semaglutide arm got there against 26% of the liraglutide arm. At 15%, it is about 56% against 12% — a fourfold gap. Liraglutide's performance at the 15% threshold (14 of 117) is closer to placebo (5 of 78) than it is to semaglutide.
And it achieves that with seven times as many injections: 365 a year against 52.
The part that was not predictable
A weaker drug is usually assumed to be a gentler one. That assumption is what makes liraglutide sound like the cautious choice, and STEP 8 does not support it.
Serious adverse events were reported in 14 of 127 liraglutide participants (11.0%), against 10 of 126 on semaglutide (7.9%) and 6 of 85 on placebo (7.1%).
The daily injection producing less than half the weight loss recorded more serious events than the weekly one — and semaglutide's rate sat within a percentage point of placebo.
Non-serious adverse events were, meanwhile, all but identical: 115 of 126 on semaglutide and 115 of 127 on liraglutide, against 68 of 85 on placebo. Around 91% of participants in both active arms reported something.
Put together, the two rows say something a mean weight change cannot. The day-to-day burden of the two drugs was the same. The benefit was not. These are single-trial figures with small event counts — 14 against 10 is a difference of four people, and no reader should treat the serious-event ordering as settled — but they point away from, not toward, the idea that the older molecule is the safer bet.
What the labels add
Two further facts sit outside the trial and complicate the picture in the same direction.
The first is gallbladder risk, where the ordering inverts what most people expect. The Saxenda label reports cholelithiasis in 2.2% of liraglutide-treated adults against 0.8% on placebo, and adds that most of those patients required surgical removal of the gallbladder. The Wegovy label reports 1.6% against 0.7% for semaglutide. The molecule producing far less weight loss carries the larger gallstone excess, and both labels state that the excess persisted after accounting for the degree of weight loss. Our gallbladder page sets this out across three products, where the pattern turns out to be consistent rather than a quirk of these two.
The second is duration. Semaglutide's elimination half-life is about a week; liraglutide is dosed daily because its is far shorter. That is why one is injected weekly and the other daily, and it is also why the Wegovy label instructs discontinuation at least two months before a planned pregnancy while the Saxenda label carries no equivalent waiting period. A longer half-life is what makes weekly dosing possible and what makes stopping slower.
Where liraglutide still has a case
None of the above makes liraglutide obsolete, and a comparison that ended there would be overstating a single trial.
Liraglutide has been approved for longer than any other molecule in this class, which means the longest post-approval safety record. It is available generically in some markets where semaglutide is not, and cost and supply have both been genuine constraints on semaglutide rather than theoretical ones. Individual tolerance varies, prior response to one drug is information a prescriber has and a trial does not, and what a given plan will actually pay for frequently decides the question before clinical preference gets a vote — the structure behind that is set out on our insurance coverage page.
What cannot be claimed on the evidence in STEP 8 is that liraglutide is the gentler or safer option. It produced less than half the weight loss, with the same rate of non-serious adverse events, a numerically higher rate of serious ones, and seven times the injections.
The limitation worth stating
STEP 8 is a single trial of 338 people, sponsored by the company that manufactures both drugs — an arrangement that cuts in an interesting direction here, since the sponsor had a marketed product on each side of the comparison. It ran for 68 weeks, so it says nothing about the following decade. And the doses tested are the top approved doses of each, which is the right comparison for this question but means it reports nothing about how the molecules compare lower down.
For how semaglutide compares with the drug that has since overtaken it, see our tirzepatide and semaglutide comparison; for what happens when any of these drugs is stopped, the trials behind that are on our stopping GLP-1s page. As our editorial standards set out, nothing here is a recommendation about which medicine anyone should take.