A summary of published research, not medical advice. See the Disclaimer.
Studies: every indexed trial and review on this topic, scored
Metformin is a prescription glucose-lowering drug, taken as a tablet, used with diet and exercise in type 2 diabetes [13]. It is old and inexpensive, and lowers blood glucose mainly by reducing glucose output from the liver. It is first-line treatment for type 2 diabetes. Interest in it as a longevity drug rests on observational data in people with diabetes and on animal studies. No completed trial has tested it for aging.
Best-supported effect: Delaying type 2 diabetes in adults at high risk. Helps (strong evidence).
Most-cited weak claim: Slowing aging or extending lifespan in people without diabetes. Unknown (very weak evidence).
How we grade evidence · Last reviewed: October 2026
What you need to know
In US adults at high risk, metformin delayed type 2 diabetes, with strong evidence: in the Diabetes Prevention Program it cut new diabetes by 31% compared with placebo over an average 2.8 years, while the lifestyle program cut it by 58%. Whether it slows aging or extends lifespan in people without diabetes is unknown, because no human trial has finished and the rest is mouse and observational data. In one trial of healthy adults over 65, it reduced muscle gains from resistance training. Diarrhea was reported in 53% of people on metformin against 12% on placebo, and low or borderline vitamin B12 was more common at 13 years. The label lists severe kidney impairment (eGFR below 30) as a contraindication and warns that lactic acidosis risk rises with kidney impairment, age 65 or over and heavy alcohol use. The Diabetes Prevention Program used 850 mg twice daily.
| Effect | Verdict | What was measured | Studied in |
|---|---|---|---|
| Delayed onset of type 2 diabetes | Helps (strong evidence) Large placebo trial of 3,234 high-risk adults who did not know what they got, diagnosis as outcome |
Hard outcome | US adults with elevated glucose and overweight |
| Fewer diabetes complications and deaths | Probably helps (moderate evidence) One trial in newly diagnosed adults who knew what they got, inconsistent sub-study |
Hard outcome | Overweight adults with newly diagnosed type 2 diabetes |
Diabetes prevention: The Diabetes Prevention Program randomized 3,234 adults at high risk. Over an average 2.8 years, metformin cut new diabetes by 31% compared with placebo. The lifestyle program cut it by 58% and worked better than metformin [1]. Helps (strong evidence): large, randomized, blinded against placebo, and the outcome is the diagnosis itself.
Diabetes outcomes: In UKPDS 34, overweight people with new type 2 diabetes assigned to metformin had fewer diabetes-related outcomes and lower all-cause mortality than those on conventional treatment [2]. In the same report, adding metformin early to a sulphonylurea was linked to more diabetes-related deaths, which a combined analysis did not confirm [2]. Probably helps (moderate evidence): downgraded for one open-label trial with an inconsistent sub-study.
| Claim | Verdict | What was measured | Studied in |
|---|---|---|---|
| Slowing aging or extending lifespan | Unknown (very weak evidence) No completed human trial, only mouse data and studies that followed people over time |
Hard outcome | No completed human trial |
| Lower all-cause mortality in people with diabetes than without | Unknown (very weak evidence) Only studies that followed people over time, groups differed at the start, repeat attempts failed |
Hard outcome | Observational cohorts |
| Lower cancer risk | Unknown (very weak evidence) Pooled analysis of 166 non-trial studies, mostly diabetes, strong tilt toward published positive results |
Hard outcome | Observational studies, mostly people with diabetes |
| Extra benefit from exercise | May make it worse (weak evidence) One trial of 94 healthy older adults, less muscle gained |
Measured | Healthy adults over 65 |
Aging and lifespan: TAME (Targeting Aging with Metformin) was proposed in 2016 to test whether metformin delays age-related diseases in older adults without diabetes [3]. None of the sources reviewed here report TAME results. A 2019 commentary argued its design may not answer whether metformin extends healthspan in people free of chronic disease [4]. Unknown (very weak evidence): no completed human trial, only mouse and observational data.
Mortality: A 2014 UK records study reported that people with diabetes starting metformin lived longer than matched people without diabetes [5]. A 2017 meta-analysis of 53 studies reported similar findings and noted baseline differences that could bias them [6]. A 2025 review describes methodological problems in these early studies, failed replication attempts, and trials in people without diabetes, and calls the anti-aging case increasingly uncertain [7]. Unknown (very weak evidence): observational data only.
Cancer: A 2024 meta-analysis of 166 studies found lower cancer incidence among metformin users. Heterogeneity and publication bias were high, and almost all data came from people with diabetes [8]. Unknown (very weak evidence): observational data only.
Exercise: In the MASTERS trial, healthy adults over 65 did 14 weeks of resistance training with 1,700 mg/day metformin or placebo. The metformin group gained less lean mass and thigh muscle [9]. A follow-up analysis found metformin dampened muscle gene responses to training [10]. May make it worse (weak evidence): one trial of 94 people and its follow-up.
Shown in humans
Shown only in animals or cells
Theory only
| Side effect | Verdict | How often vs placebo | Source | Studied in |
|---|---|---|---|---|
| Diarrhea | May happen (weak evidence) One 29-week diabetes trial, 53% vs 12% on placebo |
53% vs 12% | [13] | |
| Nausea or vomiting | May happen (weak evidence) One 29-week diabetes trial, 26% vs 8% on placebo |
26% vs 8% | [13] | |
| Flatulence | May happen (weak evidence) One 29-week diabetes trial, 12% vs 6% on placebo |
12% vs 6% | [13] | |
| Weakness (asthenia) | May happen (weak evidence) One 29-week diabetes trial, 9% vs 6% on placebo |
9% vs 6% | [13] | |
| Indigestion | May happen (weak evidence) One 29-week diabetes trial, 7% vs 4% on placebo |
7% vs 4% | [13] | |
| Abdominal discomfort | May happen (weak evidence) One 29-week diabetes trial, 6% vs 5% on placebo |
6% vs 5% | [13] | |
| Headache | May happen (weak evidence) One 29-week diabetes trial, 6% vs 5% on placebo |
6% vs 5% | [13] | |
| Taste disorder | May happen (weak evidence) One 29-week diabetes trial, 1-5%, more than on placebo |
1-5%, more than placebo | [13] | |
| Low blood sugar on metformin alone | May happen (weak evidence) One 29-week trial, 1-5%, fewer episodes than insulin or sulphonylurea diabetes pills |
1-5%, more than placebo; fewer episodes than with insulin or sulphonylureas | [2][13] | Overweight patients with type 2 diabetes |
| Low or borderline vitamin B12 | Probably happens (moderate evidence) Long-term follow-up of a trial with random assignment, 20.3% vs 15.6% at 13 years |
20.3% vs 15.6% at 13 years | [11][13] | |
| Anemia | Probably happens (moderate evidence) Long-term follow-up of a trial with random assignment, rate not reported |
Not reported | [11] | |
| Smaller muscle gains from resistance training | May happen (weak evidence) One trial of 94 older adults plus a follow-up analysis of muscle genes |
Not reported | [9][10] | 94 older adults |
| Weight gain | Probably no harm (moderate evidence) One trial where participants knew what they got, less gain than insulin or sulphonylurea pills |
Less weight gain than with insulin or sulphonylureas | [2] | Overweight patients with type 2 diabetes |
| Lactic acidosis in trial and cohort populations | Probably no harm (moderate evidence) 347 trials and studies that followed people over time, zero cases in either group |
0 cases in 70,490 vs 0 in 55,451 patient-years on other diabetes drugs | [19] | |
| Lactic acidosis with kidney impairment or other risk factors | Unknown (very weak evidence) Reports of single patients after the drug went on sale, mainly significant kidney problems, no rate given |
Case reports only | [13] | People with significant kidney impairment |
| Liver injury | Unknown (very weak evidence) Reports of single patients only, no rate given |
Case reports only | [13] |
The rates above come from one 29-week US trial in type 2 diabetes (141 on metformin, 145 on placebo), where diarrhea led 6% of metformin users to stop [13]. In the Diabetes Prevention Program Outcomes Study, each year of metformin use raised the odds of B12 deficiency by about 13% [11]; the label reports subnormal B12 in about 7% after 29 weeks [13].
This lists what studies and labels report. A pharmacist or prescriber can say whether it applies to a specific person.
This lists what studies and labels report. A pharmacist or prescriber can say whether it applies to a specific person.
| Combined with | What happened | Verdict | Source | Studied in |
|---|---|---|---|---|
| Insulin or sulphonylureas | Higher risk of low blood sugar; the label notes lower doses of these drugs may be needed | Unknown (very weak evidence) A warning on the drug label only, no measured rates given |
[13] | |
| Dolutegravir | Metformin exposure (AUC) rose 79% with 50 mg daily and 145% with 50 mg twice daily in 30 healthy adults; well tolerated | May happen (weak evidence) One study of 30 healthy adults who knew what they got, metformin blood exposure rose |
[20][13] | 30 healthy adults |
| Cimetidine | Single-dose study: metformin AUC 1.40 times and peak 1.61 times higher | May happen (weak evidence) Single-dose study, total metformin blood exposure 1.40 times higher |
[13] | |
| Ranolazine, vandetanib | Listed in the label as reducing metformin clearance through kidney transporters (OCT2/MATE), with possible lactic acidosis risk; no data shown | Unknown (very weak evidence) Listed on the drug label only, no data shown |
[13] | |
| Topiramate | Raised metformin AUC 1.25 times; it can also cause metabolic acidosis, and the label warns of added lactic acidosis risk | May happen (weak evidence) Measured study on the drug label, total metformin blood exposure 1.25 times higher |
[13] | |
| Other carbonic anhydrase inhibitors (zonisamide, acetazolamide, dichlorphenamide) | These drugs can cause metabolic acidosis; the label warns of added lactic acidosis risk | Unknown (very weak evidence) A warning on the drug label, no measured change in metformin blood exposure |
[13] | |
| Alcohol | The label states alcohol potentiates metformin's effect on lactate metabolism | Unknown (very weak evidence) Statement on the drug label only, no study data shown |
[13] | |
| Iodinated contrast | Intravascular contrast in metformin users has led to a sudden fall in kidney function and lactic acidosis | Unknown (very weak evidence) Drug label reports these events, no rates or study described |
[13] | |
| Drugs that raise blood glucose (thiazides and other diuretics, corticosteroids, phenothiazines, thyroid products, estrogens, oral contraceptives, phenytoin, nicotinic acid, sympathomimetics, calcium channel blockers, isoniazid) | The label warns of possible loss of glucose control | Unknown (very weak evidence) General warning on the drug label, no data for individual drugs shown |
[13] | |
| Glyburide, furosemide, nifedipine, propranolol, ibuprofen | Single-dose studies: metformin exposure changed by 0.90 to 1.16 times; metformin lowered glyburide and furosemide exposure; the label lists no dose changes | May not increase it (weak evidence) Single-dose studies, metformin blood exposure 0.90 to 1.16 times the usual level |
[13] |
These are the amounts used in the studies. They are not recommendations.
The Diabetes Prevention Program used 850 mg twice daily [1][11]. MASTERS used 1,700 mg/day for 14 weeks of training [9]. UKPDS 34 used metformin as first-line treatment in an open-label design [2].
Metformin is approved for treating type 2 diabetes. It is not approved for aging, and no regulator recognizes aging as a treatable condition; TAME was designed partly to create a path for that [3]. A 2021 critical review concluded that evidence that metformin extends human lifespan remains controversial [12].