Ozempic's drug gave old female mice about three extra months

In one lab, 40 old female mice given daily semaglutide lived a median of 834 days, against 742 for 39 untreated mice. Treated mice in a companion group ate about a quarter less food, and the lifespan test had no group that simply ate less, so it can't say whether the drug did more than a diet would. Nothing was tested in people.

13 October 2026 · How we rate evidence

Every day for the rest of their lives, 40 elderly female mice in a Berkeley lab got a small injection of semaglutide, the drug in Ozempic and Wegovy. Another 39 got salt water. Past day 800, 25 of the treated mice were still alive, against 10 of the untreated ones. The last untreated mouse died at 884 days; seven treated mice were still going at 900.

The claim going around. The study appeared in Nature on 2 September. Ten days later ScienceDaily posted an NIH release under the headline "Scientists find Ozempic may slow aging itself." Its summary says the drug's effects "went beyond those seen with calorie restriction." Inc. went with "Semaglutide Can Help Us Live Longer—And Not Just by Eating Less," and Morning Overview with "Semaglutide extended older mice's lives beyond what calorie restriction achieved." The paper's own title, "Late-life semaglutide treatment slows ageing and extends lifespan in female mice," keeps the mice in.

What they did. Danica Chen's team at the University of California, Berkeley, started with female mice 20 months old, well into old age for a lab mouse. Semaglutide works partly by curbing appetite, and here the treated mice ate about 24% less. That matters, because eating less, known as calorie restriction, is the best-established way to lengthen life in lab animals.

The team ran three separate experiments. In the first, mice got semaglutide or saline until they died. In the second, a new set of mice was treated for three months and then tested on strength, balance, memory and glucose control, and their tissues were examined for markers the field calls hallmarks of ageing, such as inflammation and worn-out stem cells. The third, added after reviewers asked for it, pitted semaglutide against a diet: ten mice on the drug, ten fed 24% less food without it, ten untreated, all tested at the start and after two and four months.

What they found. The lifespan result is the strongest thing in the paper. Median lifespan, the age by which half the mice had died, was 742 days for untreated mice and 834 days with semaglutide: 92 days, or 12% longer in relative terms. The difference was very unlikely to be chance, and it showed up in each of the two batches of mice the data appear to come from, by our analysis of the published numbers. Causes of death were spread similarly in both groups. The mice were randomised, analysis was blind, and the raw numbers behind nearly every figure are published.

The function tests pointed the same way. Over four months, untreated mice lost ground on almost every test, while semaglutide mice mostly held steady:

All three groups started out alike: about 66 seconds on the rod, about 20 minutes of running. On five of the nine tests the drug and diet groups weren't clearly different, but the diet itself clearly beat no treatment on only three: the rotating rod, a hanging test and the treadmill. Semaglutide did better than the diet on the other four; the authors highlight three of them: time spent in the open centre of an arena, the memory maze and glucose tolerance.

What it doesn't show. The biggest gap is the one the coverage skipped. The lifespan experiment compared the drug only with mice eating as much as they liked. There was no group fed 24% less, so the extra 92 days could be the drug, the diet it produced, or both. The calorie-matched comparison ran for five months and measured function, not lifespan. Morning Overview's headline, that the drug extended lives "beyond what calorie restriction achieved," describes an experiment that wasn't done.

The case for effects "beyond" eating less rests on 4 of 9 comparisons, each with ten mice per group. None would survive the usual correction for testing nine things at once, by our arithmetic. The people testing the diet group knew which mice they were, and by four months the semaglutide mice were somewhat lighter and leaner than the diet mice (25.0 against 26.5 grams), so leftover weight differences could matter.

"Slows ageing" is the authors' own summary, and it's broader than any single measurement. Ageing here was gauged by lifespan, by how fast function declined over four months, and by tissue markers taken once, after three months, with no starting values. Many of those markers were gene-activity readings in five or six mice, and a few protein measures used three. And this is one strain, one sex, one dose and one lab. One reviewer thought the untreated mice died young for this strain; the authors reply that 742 days is within published ranges.

What it could mean for people. The question behind the headlines is whether people taking these drugs age more slowly. This study can't answer it. By standard body-size scaling the mouse dose works out to roughly 1.6 mg a week for a 70 kg adult, within the 0.25 to 2.4 mg weekly doses used clinically (Yang and colleagues; STEP 5). But rodents clear the drug far faster than people, with a half-life of about 7 hours in rats (Knudsen and Lau) against about a week in humans, and blood levels weren't measured, so exposure may not match. The NIH release itself says the findings "do not show that Ozempic, Wegovy, or other GLP-1 drugs can extend human lifespan," and the authors call for long-term trials in older people. The ScienceDaily headline, with Ozempic as its subject and no mice in it, goes further than that text.

Why it's still interesting. Drugs that lengthen life in normally ageing mice are rare, and a treatment started in old age that adds 12% to median lifespan is notable. Semaglutide is also already approved and widely used, which makes the question testable in people sooner than for most candidates. The diet comparison adds something new: semaglutide kept up with the diet on the tests where the diet helped, without the feeding-time swings in activity and fuel use the restricted mice showed, which the authors put down to hunger. Whether it does more than a diet is still open.

Open questions.

Would calorie-matched mice live as long? Repeat the lifespan study with a third group fed the same reduced amount, ideally spread through the day as the drug mice ate. If they live as long, semaglutide is a way to get the benefits of eating less without the hunger. If the drug group lives longer, it acts on ageing some other way, and that would change which drugs are worth testing for longevity.

Are the "beyond the diet" gains real? Rerun the comparison with more mice, blinded scoring for every group, and diet mice matched on body weight. If the memory and glucose advantages hold, they point to direct effects of the drug worth studying in their own right.

Does it hold in males and other strains? Test several doses in both sexes of a genetically mixed mouse at more than one site. Many ageing treatments work in only one sex, and people are not one inbred strain.

A note on timing. The paper came out on 2 September 2026. The ScienceDaily post followed ten days later, and coverage was still arriving three weeks on.