Foundations
The things that cut across every other section: the harms worth removing, the exposures you can actually change, the handful of numbers worth knowing, and the social ties that carry one of the largest survival signals in epidemiology. Also a filter for telling a real longevity claim from a marketed one.
Don't smoke; test your home for radon and keep indoor air clean; treat gum disease; know your apoB, your lipoprotein(a) and your resting-heart-rate trend; and invest in close relationships. Most of the evidence on this page comes from large observational cohorts and genetic studies; the randomised trials mostly test single interventions, and several of the best-known ones were null.
Topics covered in depth
Smoking and nicotine →
The largest preventable-mortality lever in modern medicine, what quitting gives back at every age, which quitting aids work, and how the picture changes once you separate the combustion from the molecule: from cigarettes through vapes and pouches to the patch.
Oral health →
Gum disease as a body-wide inflammatory state, the tongue bacteria that help supply the body's nitric oxide, why the strongest antiseptic mouthwashes cut both ways, and why the heart link looks associational while the diabetes benefit is real and the pneumonia benefit probable.
Environmental toxins → · Iron and aging →
The external exposome and the internal one: which chemicals clear in days and which linger for years, what actually lowers the load, and why a metal you cannot excrete accumulates into a modifiable risk.
Cholesterol and lipoproteins →
What ages arteries is not one panel reading but the lifetime number of cholesterol-carrying particles they have been exposed to, and why apparent paradoxes usually aren't.
Resting heart rate →
The cheapest measurement on this site and its clearest example of a marker that is not a lever: what a million people's worth of cohort data shows, and what happened in the trials that lowered it on purpose.
Heart rate variability →
The most trackable physiological aging proxy, what it does and doesn't tell you, and which wearables actually measure it.
Purpose and social connection →
How strong the link with dementia and mortality really is once health and mood are accounted for, and why the effect is much harder to prove than to observe.
Blue Zones →
What survives scrutiny: Sardinia's age records held up (in a defence with disclosed conflicts), Okinawa's and Nicoya's advantages collapsed within a generation (Okinawa no longer qualifies), and Loma Linda's 4–7 year advantage at age 30 comes from Adventist cohorts, not the town.
Smoking: the biggest lever by far
Not smoking is the largest single lever on this page, by a wide margin. Tobacco kills more than 7 million people a year worldwide, including over 1.6 million non-smokers exposed to second-hand smoke, and combustion is the part that matters rather than the nicotine molecule.[1]
Nicotine outside a quit attempt is still not a safe baseline. Vaping raises heart rate and blood pressure in the short term, with no long-term outcome data yet; oral pouches are essentially unstudied; and in pregnancy every tobacco product, e-cigarettes included, should be stopped. See Smoking and nicotine.
The air you breathe and the metals you can't excrete
Air pollution is the largest environmental killer, the second-leading risk factor for death worldwide, with no safe floor. In pooled European cohorts, each 5 micrograms of fine particulate matter per cubic metre of long-term exposure tracked with about 13% higher death rates from natural causes, at the high end of published estimates (harmonised re-analyses land nearer 3–4%), and the association held, even steepened, below 12 µg/m³.[2] In US Medicare data, no threshold was detectable down to 4 µg/m³.[3] Pooling 21 studies of some 24 million people, the same increment tracked with about 8% higher dementia incidence, with certainty limited by disagreement between studies and by socioeconomic confounding.[4]
Radon is the cheapest high-yield test on this page. It is the second leading cause of lung cancer after smoking, implicated in roughly 21,000 US deaths a year on the US Environmental Protection Agency's modelling.[5] Pooled European studies found about 16% higher lung-cancer risk per 100 becquerels per cubic metre of long-term exposure, with no threshold below which it stops mattering.[6] A test kit costs about what a restaurant meal does, and a bad result has a cheap engineering fix.
Lead, cadmium and arsenic are real cardiovascular risk factors. A 2023 American Heart Association scientific statement reviewed them as contributors to heart disease, though the evidence is uneven between metals and the statement is a review, not a formal classification. Lead and cadmium stay in the body for years to decades; arsenic clears within days but usually arrives every day through water and rice. Either way, the lever is preventing intake.[7] Pulling them out afterwards does not protect the heart, and chelation belongs only to diagnosed metal poisoning: in a 1,000-patient randomised trial in heart-attack survivors with diabetes, EDTA chelation more than halved blood lead and still did not reduce cardiovascular events,[8] and chelation carries real harms: low blood calcium, kidney injury, and deaths reported to US health authorities.[9]
Cutting everyday chemical exposure moves the biomarkers. In a randomised trial, food produced and served with minimal plastic contact, plus plastic-free kitchenware, cut urinary bisphenol A by 59.7% and two phthalate breakdown products by 53.5% and 37.5%.[10] These are exposure markers, not health outcomes.
Iron is the internal version. The body has no route to excrete it, so high-normal stores are a risk signal, and supplementing without a documented deficiency is a mistake. Genetic studies link higher lifetime iron to a somewhat shorter life, but the treatment case is weak: a trial of lowering iron by blood removal was null, and both brain-iron-lowering trials made outcomes worse. The reverse error is at least as common: anaemia affects about one in ten people over 65 and more than one in five past 85, and in an older adult low iron is a reason for a work-up, since it can flag bleeding or cancer, not for self-treatment. See Iron and aging.
Your mouth: the diabetes payoff is real, the heart payoff isn't shown
Treating gum disease pays off most clearly in diabetes. In people with diabetes it lowers long-term blood sugar by about 0.4 percentage points of HbA1c (4.7 mmol/mol) at three to four months, which the review authors judge clinically significant; that is pooled across randomised trials at moderate certainty, and the single largest trial found no benefit.[11]
In randomised trials, professional gum cleaning lowered C-reactive protein, a blood marker of inflammation, by about 0.69 mg/L at six months, most in people starting above 3 mg/L.[12] The landmark vascular trial found artery function transiently worsened for 24 hours before improving over two to six months.[13] The link between gum disease and heart disease is strong in observational data, but genetic causal analysis comes up empty and the American Heart Association's 2026 statement calls causal evidence elusive.
Mouth care probably cuts hospital pneumonia: in a 2026 trial across three Australian hospitals (8,870 adult patients), enhanced mouth care cut hospital-acquired pneumonia from 1.0% to 0.7%,[14] though in nursing homes the best review could not show fewer cases.[15] See Oral health.
How much is inherited, and why it matters less than it looks
Estimates of how much of the variation in lifespan is set by inherited gene variants are modest: about 20–25% in twin studies, and as low as 6% in some large pedigree studies. A 2026 reanalysis argues the twin estimates are deflated by deaths from accidents and infections, and that the heritability of intrinsic ageing is above 50% once those are stripped out.[16] The honest range is wide, and this site does not pick a number.
The practical conclusion survives either estimate. In a large randomised trial of a multi-part lifestyle programme for dementia, the benefit was statistically indistinguishable between carriers and non-carriers of the APOE ε4 variant that raises Alzheimer's risk; because carriers start from higher risk, the usual inference is that they stand to gain more in absolute terms, an assumption the data are consistent with rather than something the trial showed. See Dementia prevention. Behaviour dominates the advice here because it is the part you can change, not because genes are negligible.
The non-genetic share has two large parts: lifestyle (what you eat, how you move and sleep, who you spend time with) and the exposome, the cumulative input from air, water, household chemicals and psychosocial stress. Higher blood cadmium, and to a lesser extent lead, goes with faster-running epigenetic clocks in survey data, though the cadmium link is largely confined to smokers. See Environmental toxins.
The numbers worth knowing
Resting heart rate is the cheapest and most replicated. Across more than a million people, mortality rises roughly 9–17% for every 10 beats per minute, and the link survives adjustment for directly measured fitness. It is a marker, not a lever: lowering it with a drug helps in heart failure, does nothing in coronary disease without heart failure, and has never been tried in healthy adults. Track your own trend, not a single reading. See Resting heart rate.
Heart rate variability, the millisecond fluctuation between heartbeats, is a continuously trackable read on how well the nervous system regulates the body. It declines with age and tends to be lower with inflammation or poor metabolic health, though those links are observational and looser than often claimed; whether raising it extends life is unproven, and wearables vary widely in how well they measure it. See Heart rate variability.
ApoB is the one lipid number worth learning. Lifetime exposure to apolipoprotein B particles causes atherosclerosis, and apoB counts those particles directly rather than the cholesterol inside them; whether everyone needs it measured, rather than people whose standard panel is ambiguous, is still argued. See Cholesterol and lipoproteins and Lipid management.
Lipoprotein(a) is the test to run once. About one in five people inherit a high level; it is set genetically and essentially unmoved by diet, exercise, weight loss or statins, so a single measurement tells you whether you carry it. If it is high, push harder on everything that is modifiable.
Composite scores are a checklist, not a precision instrument. The American Heart Association's Life's Essential 8 scores diet, physical activity, nicotine, sleep, body mass index, blood lipids, blood glucose and blood pressure, each 0–100.[17] In a long-running US cohort, adults above the median had roughly 45% lower mortality over three decades and about half the rate of cardiovascular events.[18] But it predicts mortality no better than the simpler seven-component version it replaced, and adds only marginal accuracy to a conventional risk model.[19]
For epigenetic clocks and the rest of the biological-age apparatus, see The Biology of Aging.
What to work on, in order
A reasonable order for healthy midlife adults:
- Remove the documented harms first. Smoking at zero, alcohol as near it as you can manage, sleep apnea screened rather than assumed, long motionless stretches broken up.
- Build cardiorespiratory fitness and strength. Fitness is the most replicated predictor of the lot: each one-unit rise (one metabolic equivalent, 3.5 mL of oxygen per kilogram per minute) tracks with 11–17% lower all-cause mortality across 199 cohorts, though the data are observational throughout and the authors grade certainty very low to moderate.[20] Meeting both the aerobic and the strength guidelines tracks with about 40% lower mortality, against 29% for aerobic work alone and 11% for strength alone.[21] See Exercise.
- Eat a vegetable- and fish-heavy pattern (Mediterranean, MIND), optionally with an eating window that ends before late evening. See Nutrition.
- Keep sleep regular. In 60,977 UK Biobank adults with a week of wrist-tracker data, the most regular fifth had 48% lower all-cause mortality than the least regular, falling to 30% once income, deprivation, smoking, activity, shift work and blood-pressure or cholesterol medication were accounted for.[22] See Sleep.
- Invest in social connection. A 148-study meta-analysis puts strong relationships at roughly 50% higher survival odds, though newer and larger data on isolation and loneliness find smaller effects, and genetic analyses find little evidence that isolation itself causes the diseases it tracks with (observational); it is also the strongest single late-life protective marker for dementia, roughly 30% lower risk against about 9% for cognitive activity. See Cognitive engagement. Purpose travels with it, with a much wider hedge: the association is modest (on the order of 15–25% lower risk, shrinking once health and mood are accounted for), some of it runs backwards, since declining health drains purpose, and no trial has shown that raising purpose extends life. See Purpose.
- Clean up the air and water you live in. A bedroom air filter, a radon test, a vented extractor over the stove, fewer scented products, decent water filtration, glass and steel for food. Individually small and cheap.
- Heat and cold if you enjoy them. Sauna's large mortality association comes almost entirely from one Finnish cohort (the only large independent cohort tested dementia, not mortality, and found about a third of the effect), and twenty pooled randomised trials of passive heating found nothing on artery function, glucose, lipids or inflammation; blood pressure moved only with whole-body heating and in people with cardiovascular risk. Cold exposure has less evidence still and real risks: drowning is the leading cause of cold-water immersion death, and arrhythmia, cold urticaria and pregnancy are reasons to avoid it. See Sauna and Cold exposure.
The pharmacological frontier sits below all of this for healthy adults and is thinner than its coverage suggests. TAME, the metformin trial usually cited as the field's flagship, has never enrolled anyone and has only a fraction of the $45–70 million it needs. PEARL, the largest and longest trial to date of intermittent rapamycin in healthy adults, missed its primary endpoint and used a compounded drug at about a third of the generic's bioavailability. Senolytics, drugs that clear worn-out cells, accelerated several epigenetic clocks in a 19-person study of healthy volunteers; the effect mostly faded and never appeared on the two best-validated clocks, but it is reason enough for healthy adults to stay away. See Geroprotectors and Clinical Care.
For the full ordered action list drawn from every pillar, see the healthspan long list.
How to evaluate a "longevity intervention"
Five questions for any new claim:
- What kind of evidence is it? Animal data alone transfers poorly to people. A biomarker moved in humans is suggestive, not proof. A randomised trial with a hard endpoint is the standard. Genetic (Mendelian randomisation) studies, which use inherited variants as a natural experiment, support strong causal inference and often deflate observational claims.
- Who paid for it? A large share of the positive longevity-supplement literature rests on short industry-funded trials.
- Is the population relevant? A signal in 80-year-olds with muscle wasting may not carry over to a healthy 45-year-old.
- What is the absolute effect? A "30% relative risk reduction" of a rare event is small in absolute terms.
- What is the alternative? Often the best move is removing a harm (smoking, ultra-processed food, untreated sleep apnea), not adding something new.
What's overrated
- Single-nutrient and single-molecule fixes. Patterns of eating, moving and sleeping across years consistently beat them.
- "Natural" as a safety claim. Iron, calcium and high-dose vitamins are natural, and all carry documented harm above a threshold.
- Detoxing. Chelation outside diagnosed poisoning failed its definitive trial (above); "detox" cleanses, juice fasts and anti-toxin supplement stacks have no outcome evidence in healthy adults. Prevent exposures at the source.
- Air-purifier theatre. Houseplants would need something like 10 to 1,000 per square metre of floor to match a home's ordinary air exchange, and oxidation-based purifiers (ionizer, plasma, ultraviolet, photocatalytic) generate secondary pollutants; in one standardised test, some put out more volatile organic compounds than they removed. A plain HEPA filter and an open window work.
- Daily mouthwash for a healthy mouth. Chlorhexidine and some other antibacterial rinses strip the tongue bacteria that help make the body's nitric oxide, with no matching benefit. Brushing and cleaning between the teeth are the intervention.
- Aggressive phlebotomy as anti-aging therapy. Offloading genuine overload is sensible; bleeding a healthy person toward deficiency is not, and the one large randomised test was null.
- Blue Zone mimicry as a programme. The behaviours hold up on their own; the certification business and the supercentenarian records do not, and the demography is fragile at the 110-plus tail rather than across every age claim.
- Reading a single cholesterol panel as a verdict. LDL cholesterol and apoB disagree in roughly 5–25% of people, and when they do, risk follows apoB.
- Optimising markers you cannot act on. A number is worth measuring when a specific result leads to a specific different decision.