On 2 September 2026, a team led by Danica Chen at UC Berkeley’s Department of Metabolic Biology & Nutrition reported in Nature that starting semaglutide — the GLP-1 receptor agonist sold as Ozempic and Wegovy — in old age extended median lifespan in female mice from 742 to 834 days, about 12%. Berkeley’s news office put the average gain at roughly 100 days. The mice were already 20 months old when treatment began — roughly a human woman in her 60s — and only females were studied. This is not a human longevity trial, and it does not prove that Ozempic extends human lifespan.

Lead author Yufan Feng and co-authors from the University of Copenhagen and the U.S. National Institute on Aging (NIA) frame the result as late-life GLP-1 activation slowing ageing and acting as a calorie-restriction mimetic — with extras that diet alone did not match in their head-to-head tests.

Why it matters

Calorie restriction without malnutrition is one of the most reliable ways to slow ageing in laboratory animals. It is also hard for people to keep for decades. Longevity labs have hunted for drugs that capture those benefits without the hunger. Semaglutide already cuts appetite; Chen’s group asked whether that made it a practical calorie-restriction mimetic — and whether starting late still paid off.

The stake for a stranger is concrete. Semaglutide is already prescribed for type 2 diabetes and obesity, with approved or trial-backed benefits for heart, kidney, and liver disease. If a drug that millions already take also blunts hallmarks of ageing in old animals, that raises a measurable research question: how much of its wide benefits come from slowing ageing itself? The paper does not answer that for humans. It does put a named, widely used molecule on the longevity-mimetic map with a published survival curve.

Chen told UC Berkeley’s Metabolic Biology & Nutrition desk the work “could potentially broaden the application of GLP-1 medicines to healthy aging individuals to extend lifespan and healthspan.” In the same release, she stressed there is “no indication that similar results could be immediately achieved in humans.” Outside the lab, Tim Rhoads of the University of Wisconsin–Madison told Chemical & Engineering News that semaglutide is “probably the best caloric-restriction mimetic I have seen” on the data so far — then flagged the need to test males and watch muscle loss. Nir Barzilai of Albert Einstein College of Medicine told Scientific American he still wants a direct lifespan comparison of drug versus diet; Chen said that experiment is ongoing.

Key numbers

MetricValueSource
Publication2 Sep 2026, NatureDOI 10.1038/s41586-026-10940-7
InstitutionsUC Berkeley (lead) + Copenhagen + NIANature; Berkeley MBN
Strain / sexFemale C57BL/6Nature
Age at start20 months (~human 60s)Nature; SciAm; C&EN
Median lifespan742 → 834 days (~12%)Nature Fig. 1a; SciAm
Average lifespan gain~100 days longerBerkeley MBN
Post-start median survival4.7 vs 7.8 months (saline vs drug)C&EN
Lifespan cohort39 saline, 40 semaglutideNature Methods
Dose10 nmol/kg daily subcutaneousNature Methods
Food-intake cut~24% (drug); CR matched at 24%Nature

Do not translate “12%” or “100 days” into human years. Mouse median lifespan and human life expectancy are not interchangeable clocks, and this protocol was never run in people.

How they showed it

Lifespan. Twenty-month-old female C57BL/6 mice from the NIA received daily subcutaneous saline or semaglutide for the rest of life. Median survival rose from 742 to 834 days. C&EN reported the post-start medians as 4.7 months for saline and 7.8 months for semaglutide. Endpoint categories did not differ significantly between groups; age at death looked delayed across several non-tumour categories, consistent with a broad delay in physiological decline rather than a single-cause fix.

Three-month healthspan arm. A separate cohort got the same daily regimen for three months. Treated mice moved and explored more, scored better on Barnes-maze spatial memory, lasted longer on rotarod and inverted-screen tests, ran farther on a treadmill, and cleared glucose more effectively. Several motor gains still held after adjusting for body weight. Weight loss came mainly from fat; lean-mass percentage rose.

Ageing hallmarks. The same arm showed less inflammation and cellular senescence, fewer DNA-damage marks, better mitochondrial and proteostasis readouts, and healthier haematopoietic stem-cell profiles. In the dentate gyrus, BrdU+ and doublecortin+ counts rose — signs of more neurogenesis. Liver RNA-seq pointed at quieter inflammatory and lipid programs and stronger adaptive-immune, insulin-response, and proteostasis pathways — overlapping calorie restriction. NAD+ rose, several sirtuins were induced, and circulating IGF1 fell.

Head-to-head with diet. In a five-month comparison, one group got semaglutide, another a 24% calorie cut matched to the drug’s appetite effect, and controls ate freely. Both cut body weight and fat. Semaglutide matched calorie restriction on several motor measures. On exploratory drive, spatial memory, and glucose control, the drug improved animals above baseline and often beat the diet arm’s trajectory. Feeding patterns differed: restricted mice gobbled their ration then fasted; drug-treated mice ate more gradually without the same hunger-foraging spike. Chen’s group reads that as evidence the drug is more than “eat less,” even though intake fell by the same fraction.

What this is not

  • Not a human longevity trial. No older adults were randomized to semaglutide for lifespan. Human approvals for heart, kidney, and metabolic disease do not equal a proven life-extension indication.
  • Not proof that Ozempic or Wegovy extend human lifespan. Brand names appear because that is the same molecule; the survival curve is for female mice under a defined 10 nmol/kg daily injection protocol.
  • Not both sexes. Females only — chosen to limit male aggression and injury in long group-housing studies. Male results are unknown.
  • Not a finished diet-versus-drug lifespan race. The five-month arm compared function, not full lifespan. That head-to-head survival experiment is still in progress.
  • Not a free pass on muscle. Outside experts told C&EN that GLP-1–related lean-mass loss remains a clinical worry.
  • Not a claim that ageing is already a regulatory indication. Dedicated human longevity trials for GLP-1 drugs remain few and small.

The University of California has filed a patent application on GLP-1 receptor agonists for healthy ageing (64/113,481, per the paper’s competing-interests statement). That is disclosure, not a product.

What to watch

  1. Male mouse replication of the lifespan and hallmark results under the same late-life start.
  2. Chen’s ongoing lifespan comparison of semaglutide versus matched calorie restriction.
  3. Whether dual agonists (for example tirzepatide) show larger or different ageing effects in animals and in small human longevity-oriented studies already underway elsewhere.
  4. Human work that measures ageing biomarkers, muscle retention, and hard clinical outcomes in older adults — without overselling mouse medians as years of human life.

Old female mice on a daily GLP-1 shot lived longer, moved better, and in several cognitive and metabolic tests outpaced a matched diet. The number that stuck — 742 to 834 days, about 12% — is real, published, and still a mouse number. The next evidence has to come from males, from the missing diet-versus-drug lifespan race, and eventually from people. Until then, the honest headline stops at the cage.

Sources