Metformin has been in use since the 1950s and is the most widely prescribed type 2 diabetes drug in the world. It works by reducing the liver's glucose output and improving cells' sensitivity to insulin — cheap, generic, and backed by decades of safety data.
Its rise in longevity circles traces back to the 2010s. The drug activates AMPK, the cell's energy-sensing pathway, which overlaps with some of the cellular responses triggered by caloric restriction — enough to push researchers to ask whether metformin might delay aging itself. In 2015, the FDA approved metformin for testing in a trial aimed directly at age-related disease, the first time a drug was cleared to be tested against an "anti-aging" target. Headlines followed quickly, including some suggesting the drug could extend human lifespan toward 120 years.
So where did that promise land, a decade on? The answer is considerably messier than the early headlines suggested.
A 2013 study in Nature Communications reported that giving middle-aged male mice a low dose of metformin (0.1% w/w in their diet) extended both healthspan and lifespan, while a higher dose (1%) turned out to be toxic. In that study, Martín-Montalvo and colleagues found metformin mimicked some effects of caloric restriction — better physical performance and insulin sensitivity — without reducing food intake; the proposed mechanism involved AMPK activation along with reduced oxidative damage and chronic inflammation.
That positive signal became the standard reference point for the metformin-longevity link for over a decade. But the picture shifted in 2025. A large meta-analysis we also mention in our article on rapamycin, published by Ivimey-Cook and colleagues in Aging Cell, pooled 911 effect sizes from 167 papers across 8 vertebrate species, from fish to primates. While caloric restriction robustly and consistently extended lifespan, and rapamycin did nearly as well across species, metformin showed no significant benefit.
That doesn't invalidate the 2013 finding — it's still a real result from a real experiment. But the fuller picture is more complicated: metformin can show a benefit under specific conditions (low dose, starting in middle age, in one species and one experimental group), while a broad review finds no consistent signal across species.
On the human side, the most cited source for the metformin-longevity story is a 2014 UK study published in Diabetes, Obesity and Metabolism. Bannister and colleagues used the UK's primary-care database (CPRD) to compare 78,241 people started on metformin monotherapy, 12,222 started on a sulphonylurea, and 90,463 matched people without diabetes. The result was striking: people with diabetes on metformin survived longer than their matched, non-diabetic counterparts. Headlines quickly turned it into "people with type 2 diabetes can live longer than those without."
That finding was tested eight years later. In 2022, Keys and colleagues used Danish national health registers (1996-2012) to compare both singleton individuals and twin pairs discordant for metformin use — a twin design that controls for genetics and childhood environment far more tightly than Bannister's study could. The result was the opposite of Bannister's: in unadjusted analyses, people started on metformin had 52% higher mortality than non-diabetic controls among singletons, and 90% higher among discordant twin pairs. Adjusting for comorbidities and social factors shrank those figures (to roughly 32% and 64% respectively), but the direction didn't change: metformin users were dying sooner, not later, than people without diabetes.
The discrepancy likely comes down to method. Bannister's design may have dropped people early once their disease progressed and they switched to other medications — effectively filtering the metformin group down to a healthier subset over time. The twin design used by Keys and colleagues is far more resistant to that kind of bias. The "diabetics on metformin outlive non-diabetics" claim still circulating in longevity circles most likely traces back to that first, methodologically flawed study.
Part of why the observational studies disagree so sharply is simple: neither one is randomized. In observational data, whatever determines who takes metformin is already tangled up with a disease state, which makes it very hard to draw a reliable conclusion about cause and effect. The tool to fix this is well known: TAME (Targeting Aging with Metformin), a double-blind trial planned to randomize roughly 3,000 non-diabetic adults aged 65 to 79 to metformin or placebo over six years.
The FDA formally approved the trial's design in 2015, along with a composite primary endpoint covering heart attack, heart failure, stroke, most cancers, cognitive decline/dementia, and death — a decision that made headlines at the time as the first instance of a drug being tested against an aging target directly. But more than a decade after that approval, the trial still hasn't launched; the main obstacle is funding. Its estimated cost runs to around $60 million, and reporting from early 2024 indicated the project remained only partially funded. The Saudi-backed Hevolution Foundation has reportedly been considering contributing since 2022, but there's no public confirmation that this has been finalized.
It's a somewhat ironic corner of the longevity field: arguably the cheapest, safest "longevity drug" candidate with the most hope riding on it still hasn't managed to launch the one trial that would settle the question, a decade on.
A less-discussed but practically important side of the metformin story is its interaction with exercise. In a 2019 randomized trial in Aging Cell, Konopka and colleagues assigned 53 older adults to placebo or metformin and put both groups through the same 12-week aerobic training program. Exercise produced benefits in both groups, but the metformin group gained less in VO2max and whole-body insulin sensitivity than the placebo group, and the exercise-driven increase in skeletal muscle mitochondrial respiration was blunted as well.
Read alongside the fitness-mortality relationship we cover in our VO2max article, this creates an interesting tension: metformin is meant to improve metabolic health, yet it may dull some of the adaptive response that regular aerobic training produces. This is one small trial (53 people), and it doesn't support a blanket "don't exercise on metformin" conclusion. But for anyone training consistently and hoping to get the most out of both the metabolic and cardiorespiratory benefits of exercise, it's a worthwhile interaction to flag with a doctor.
For someone diagnosed with type 2 diabetes or prediabetes, the metformin question sits in a different place entirely: there, it's an already-proven, widely used treatment for blood sugar control, decided with a doctor and outside of the longevity debate.
The trouble starts with people who don't have diabetes and are seeking out metformin purely to "slow aging." The evidence doesn't support that use: a consistent lifespan benefit in animals wasn't confirmed by a broad meta-analysis, the human observational data contradicts itself, and the randomized trial that would settle the question still hasn't started. On top of that, metformin has known side effects — gastrointestinal issues like nausea and diarrhea are common, long-term use can impair B12 absorption, and lactic acidosis, while rare, is a serious risk worth keeping in mind, especially for people with impaired kidney function.
Sourcing metformin without a prescription to use "preventively" isn't advisable for someone without diabetes. If metformin is something you want to pursue, it should be a decision made with a doctor and followed with regular bloodwork.
Some older animal studies, such as a 2013 Nature Communications paper using low-dose metformin in mice, found an extended lifespan. But a comprehensive 2025 meta-analysis, pooling data from 167 papers, found no significant lifespan benefit from metformin across 8 vertebrate species. No randomized human trial (TAME) designed to test this directly has started yet.
A famous 2014 UK study (Bannister et al.) suggested exactly that. But a 2022 replication using Danish registry and twin data (Keys et al.) found the opposite: diabetics on metformin had higher mortality than matched non-diabetic controls. The original finding is now thought to likely reflect selection bias.
TAME (Targeting Aging with Metformin) is a large randomized trial, designed to directly test whether metformin delays age-related disease, whose design the FDA approved back in 2015. But the project still hasn't launched after more than a decade, mainly for lack of full funding; the most recent public reporting, from early 2024, says it remains only partially funded.
A randomized trial in older adults (Konopka et al., 2019) found that during 12 weeks of aerobic training, people taking metformin gained less in VO2max and insulin sensitivity than those on placebo. It's a single, small trial, but it's a relevant interaction for anyone training regularly and hoping to maximize exercise's metabolic benefits.
"Metformin might be the clearest example in longevity science of a drug we've had, cheap and safe, for decades — and we still can't answer the simplest question about it: does it actually extend life?"
— Vivelong Longevity Science Team