Why Some People Live Longer Than Others — And What That Means For Your Own Lifespan.

The 20% rule that ran the field for thirty years

Ask anyone in longevity research what fraction of human lifespan is genetic and you will get back the same answer: 20 to 25 percent. That number comes from twin studies — comparing identical twins to fraternal twins to separate genetic contribution from shared environment — and it has been remarkably consistent across cohorts and decades. The Danish Twin Registry, which holds health and survival data on more than 180,000 twins born since 1870, has been the primary engine of this consensus.^[2] Pedigree studies have pushed the estimate even lower, with some analyses suggesting heritability as low as 6 to 7 percent once shared environment is more rigorously controlled.^[1]

The implication that followed felt clean: most of how long you live is up to you. The 70 percent that wasn't genes had to be lifestyle, environment, healthcare access, luck. The whole longevity-as-lifestyle frame — sleep, food, training, stress, relationships, don't smoke, don't crash a motorcycle — is downstream of those twin numbers. Longevity as a field has spent two decades writing protocols on top of a finding that said you were mostly driving the car.

That finding was probably wrong. Or more accurately — the way we calculated it was hiding the actual signal.

The 2026 correction — what doubled the number, and what it doesn't say

A 2026 paper in Science by Shenhar and colleagues argues that twin and pedigree estimates have been systematically depressed by a single confound: extrinsic mortality.^[1] Twin pairs that should look more correlated in their lifespans don't, because one of them died in a car accident at 32, or from sepsis at 41, or from a war at 19. None of those deaths reflect the underlying biological aging machinery the studies were trying to measure. They are noise in the lifespan signal — and once you model them out using twin cohorts raised together and apart, the heritability of intrinsic lifespan moves above 50 percent.

That is a large correction. It moves human lifespan into the same heritability range as most other complex human traits — height, IQ, personality dimensions — and it brings the estimate in line with what we already see in other species. It also forces a reframe of three decades of population-health messaging. Genes are doing more of the work than the 20 percent number suggested.

Two things this paper does not say, both of which matter:

• It is one study. The Shenhar correction is published, peer-reviewed, and in a top-tier journal — but it is a single methodological argument applied to existing twin data, not a confirmatory replication across multiple independent cohorts with the same correction applied. The traditional 20 to 25 percent estimate has decades of replication behind it across countries and registries. The new estimate has months. Conservative position: treat the corrected number as the most plausible upper bound for now, and the uncorrected number as the most plausible lower bound. The truth is in that range. We will know more in three years.
• It does not say lifestyle stops mattering. Heritability is a population-level statistic. It tells you how much variation in lifespan within a population is explained by genetic variation within that population — not whether lifestyle interventions work for the individual sitting in front of you. Smoking, untreated hypertension, sedentary behavior, terrible sleep, and a poor diet still shorten the lifespan of the person who is doing them, regardless of what the heritability number is. Heritability tells you about variance. It does not tell you about cause.

Heritability is a statistic about variance, not a verdict about your life. Genes set the slope. Lifestyle sets where on the slope you end up.

Centenarians are a different question entirely

Average lifespan and exceptional lifespan are not the same biology, and treating them as if they are has caused most of the confusion in longevity reporting. Reaching average life expectancy — about 81 in Canada, 79 in the United States — is a story dominated by avoiding the things that kill people in their 50s and 60s: cardiovascular disease, type 2 diabetes, cancer, accident. The genetic signal for that is modest. The lifestyle signal is strong.

Reaching 100, and especially reaching 105 or 110, is a different signal entirely. Centenarians appear in families. They cluster. Their siblings live longer. Their children live longer. The heritability component of exceptional longevity is far larger than the heritability of ordinary lifespan, and the genetic variants implicated have been replicated across multiple populations — most consistently the APOE epsilon-2 allele (a variant of the apolipoprotein E gene that confers cardiovascular and Alzheimer's protection) and FOXO3, a transcription factor central to insulin/IGF-1 signaling and stress resistance.^[3][4]

A 2021 whole-genome sequencing study of 81 Italian semi-supercentenarians and supercentenarians (ages 105+/110+) added a clearer mechanistic story: the people who reach those ages carry a peculiar genetic background associated with efficient DNA repair, and they show a lower somatic mutation load than younger controls. They are not just lucky. They are running a different version of the genomic-maintenance machinery.^[5]

Centenarians also share a distinct biomarker profile that is not easily replicated by lifestyle alone: low circulating IGF-1, elevated HDL cholesterol, low NT-proBNP (a cardiac stress marker), and preserved albumin into the oldest ages.^[4][12] The biomarker pattern aligns with the genetic findings — the same insulin/IGF-1 axis that animal lifespan studies have pointed to for thirty years also appears to be central to human exceptional longevity. This matters for the protocol question later: every serious longevity intervention eventually touches the GH/IGF-1 axis, because the axis is real.

The honest division: roughly 80 to 85 percent of your population's variation in lifespan is in the range where lifestyle is the dominant lever. The top 15 to 20 percent — exceptional longevity — is where the genetic signal becomes loud, and where most lifestyle interventions hit a ceiling.

Blue zones — useful frame, weak evidence

The five regions popularly identified as "Blue Zones" — Okinawa, Sardinia (Ogliastra), Nicoya, Ikaria, and Loma Linda — have done more for public interest in longevity than any other concept of the last twenty years. They also rest on weaker evidence than the popularity suggests. A 2025 scoping review in Aging and Disease examined 65 records across ten claimed Blue Zone regions and concluded that the evidence base is mixed at best. Some regions (Ogliastra, Okinawa, Nicoya) do show higher centenarian rates than national averages. Some (Loma Linda) have no published Blue Zone studies at all. The Caribbean regions (Martinique, Guadeloupe) show patterns that may reflect age-reporting errors more than true longevity.^[6]

The bigger problem is causal: even where the centenarian rates are real, the leap from "these populations live longer" to "their diet is the reason" is an ecological inference. A 2022 critical review in Maturitas made this point directly — Blue Zone diets have changed dramatically over the past century, are not internally homogeneous, and have never been tested as discrete causal interventions in a controlled setting.^[7] The Okinawan diet of the 1950s — the one associated with the centenarian cohort — looks very different from the Okinawan diet of 2026. Genetic isolation, low extrinsic mortality (no wars, low road-traffic exposure), social cohesion, and physical-activity-built-into-daily-life almost certainly contribute alongside diet, and disentangling those is impossible in observational data.

That doesn't make Blue Zone principles useless. Plant-forward eating, low ultra-processed-food intake, social connection, daily walking, moderate caloric intake — these all align with the foundational tier below, and they have independent evidence for survival benefit. They just don't have the causal strength the marketing implies.

Foundational tier — the 70% that does most of the work

Setting aside the heritability debate, the question that matters in practice is what you can actually do. The foundational tier — sleep, food, training, stress, relationships — does most of the population-level work, and the evidence for it is the strongest in the longevity space. A 2020 overview of Cochrane systematic reviews covering 187 randomized trials and 27,671 participants found that exercise reduces all-cause mortality by 13 percent.^[8] A 2025 umbrella review of 48 meta-analyses on physical activity and mortality found suggestive evidence across measured daily steps, leisure-time activity, and combined aerobic-plus-resistance training for lower all-cause and cardiovascular mortality.^[9]

A 2022 meta-analysis of behavioral lifestyle factors pooled across more than 2.8 million people found that the combination of regular physical activity, healthy diet, 7 to 8 hours of sleep, and BMI within the normal range each independently predicted greater survival in older adults.^[10] None of those four interventions is mysterious. None requires a prescription, a protocol, or a longevity clinic. They are also the four things most people don't do consistently.

Three honest framings of this tier:

• The mind plays the biggest role. Foundational health is not really about the food or the exercise — it is about the behavioral consistency that puts them in place. Stress, sleep, relationships, and mental state drive the cortisol and inflammation patterns that determine whether the rest of the protocol stack even has a chance to land. Aggressive screening anxiety, hypervigilance about disease, and a chronic foreboding mindset have their own measurable health cost. The foundational work is at least as much psychological as it is biological.
• Compliance, not protocol, is the limiting factor. The lifestyle intervention that you actually do for ten years outperforms the optimal intervention you do for three months. Every longevity protocol that ignores this fails on contact with normal human life.
• This tier covers most of the modifiable variance. If you do the four basics consistently and never run another intervention, you will capture a substantial fraction of what is actually available to you. The next two tiers are the additional 20 to 30 percent — the preservation layer on top of the foundation, not a substitute for it.

Research-curious tier — preservation, not enhancement

For users already running the foundational tier well — and only those users — the next layer is what we have started calling the preservation tier. The frame matters: this is about maintaining organ-level function while the tissue is still responsive, not about chasing biomarker scores or pushing performance. The aim is to stay out of the surgical-and-replacement bucket later, not to add a decade to a graph.

The compounds that sit in this tier — peptide bioregulators, growth-hormone-releasing peptides, glutathione precursors, mitochondrial-support compounds — share two characteristics. They preserve a normal signaling pattern rather than override it. And they have a coherent mechanism story plus enough human data to be worth considering under clinician supervision, but not enough to qualify as proven longevity interventions.

The strongest example of the preservation logic is peptide bioregulators — short tissue-specific peptides (typically two to four amino acids) characterized by Russian gerontologist Vladimir Khavinson from 1973 onward. The mechanism is counter-intuitive but clean: shorter peptide chains bind DNA promoter regions for tissue-specific genes more selectively than longer peptides do, regulating gene expression in the target organ. A 2025 review in Current Aging Science consolidates the experimental and clinical evidence across multiple compounds — Epitalon for pineal/circadian regulation, Thymalin for thymic immune function, Cardiogen for cardiac tissue, and others — into a unified picture of organ-specific preservation.^[11] A 2022 in vitro study replicated the anti-inflammatory and proliferative-modulation findings on monocytes and macrophages.^[11]

The bioregulator evidence base is real, but asymmetric. The Russian clinical tradition has used these compounds for decades with documented mortality benefit in the foundational stack (Thymalin + Epitalon in older adults). Western pharmaceutical companies have not funded confirmatory intervention trials, primarily because the compounds are off-patent peptide preparations rather than novel patentable molecules — the economics don't support the trial cost. That is not a conspiracy; it is the standard pattern for any preservation-tier compound that fixes nothing billable. Our reference page on bioregulators covers the per-compound evidence in more depth.

Growth-hormone-releasing peptides — CJC-1295, ipamorelin, sermorelin, tesamorelin — sit in the same tier with a different mechanism. They restore the pituitary's natural pulsatile GH release pattern rather than flooding the receptor the way exogenous human growth hormone (HGH) does. The pulsatility-plus-feedback design is the entire safety case: the body's homeostatic shut-off remains intact, IGF-1 doesn't stay supraphysiologically elevated for hours, and the cancer-risk signal that exogenous HGH carries is substantially reduced. Every serious longevity protocol eventually touches the GH/IGF-1 axis because the axis is mechanistically central; the choice is how you touch it, not whether.

None of this is a recommendation. The preservation tier requires clinician supervision, lab work to confirm there is something to preserve, family-history screening, and an honest assessment of which signal you are actually pulling. Users with active or recent cancer, strong hormone-sensitive cancer family history, or untreated metabolic disease are not the population for this tier. The reader who matches it knows who they are.

Experimental tier — rapamycin, senolytics, and what the human data shows

The third tier is where the most interesting science is, and also where the gap between animal data and human evidence is largest. Rapamycin — an mTOR inhibitor originally developed as a transplant immunosuppressant — extends lifespan in every mammalian model it has been tested in, often by 15 to 25 percent, and remains the most reliable pharmacological longevity intervention in the preclinical record.^[13]

The human translation is genuinely early. A 2018 pilot RCT in 25 generally healthy older adults established that low-dose rapamycin (1 mg/day for 8 weeks) was reasonably tolerated, with no significant clinical changes in metabolism, cognition, or physical performance over the short trial window.^[14] Larger and longer trials — the PEARL trial, RAP-PAC, the Mannick group's work on rapalogs for immune function in older adults — are ongoing and have produced suggestive but not confirmatory results. As of 2026, no rapamycin trial has been long enough or large enough to establish whether the lifespan or healthspan benefit observed in mice translates to humans. The mechanism is plausible; the human evidence is emerging. Our deep-dive on rapamycin in humans covers the trial data and the dosing-window debate in detail, and the rapamycin reference page tracks current status.

Senolytics — drugs that selectively clear senescent ("zombie") cells — sit at a similar stage. The dasatinib-plus-quercetin combination has the most human data, fisetin has the most ongoing trials, and the early biomarker results are encouraging but not conclusive. Our coverage of the 2025 fisetin trials walks through the current frontier.

Epigenetic reprogramming sits even further upstream. A 2021 pilot RCT in 43 men aged 50–72 showed that an 8-week diet, sleep, exercise, and supplement intervention reduced DNA methylation age (Horvath clock) by 3.23 years compared with controls.^[15] A 2024 review in Ageing Research Reviews summarizes the broader epigenetic reprogramming field, including the partial-reprogramming-via-transcription-factor approach that is currently the most aggressive bench-to-bedside frontier.^[16] The big caveat: epigenetic clocks are biomarkers, not survival endpoints, and moving the clock has not yet been shown to move actual mortality. Our reference on biological age tests covers what these clocks do and don't tell you.

Grey areas — the trade-offs, named

Three trade-offs that deserve naming directly:

• The screening trap. Aggressive disease screening is reasonable for long-term users of compounds with mechanism-plausible cancer risk (any chronic GH-axis manipulation, anything that touches IGF-1 sustainably). It is not reasonable as universal advice for healthy adults doing the foundational work. Constant screening anxiety, false-positive cascades, and incidental-finding follow-ups have their own measurable health cost. Screen aggressively for the risk profile you actually have, not for a hypothetical one.
• The cost-of-care problem. Most of the preservation and experimental tier requires lab work, clinician access, compound sourcing, and ongoing monitoring that the foundational tier does not. The marginal benefit per dollar drops sharply as you move up the tiers. A reader who would have to compromise on the foundational tier in order to fund the experimental tier is making the wrong trade.
• The mortality data gap. Almost nothing in the preservation or experimental tier has confirmed mortality data in humans. The animal data is real. The biomarker data is real. The hard endpoint — does this person actually live longer — has not been tested at scale for any of these compounds in humans, and may not be testable in the lifetime of anyone reading this. The honest user takes this tier on the strength of mechanism and biomarker evidence, not because the lifespan claim has been proved.

What we still don't know

The most consequential open questions, stated directly:

• Whether the Shenhar correction holds. The 50 percent intrinsic-heritability finding needs independent replication in different twin cohorts with different extrinsic mortality profiles before it becomes the new consensus. Until then, the field is in a useful uncertainty range — meaningfully higher than 20 percent, probably lower than 50, with the exact number to be determined.
• Whether centenarian-associated genetic variants are actionable. APOE epsilon-2 and FOXO3 are reproducibly associated with extreme longevity, but neither is currently a target for therapeutic intervention. Whether knowing your APOE genotype changes anything about how you live — beyond what general cardiovascular and dementia-risk screening already says — remains an open practical question.
• Whether any longevity intervention extends lifespan in humans. No human intervention — not rapamycin, not metformin, not senolytics, not bioregulators, not GH-releasing peptides, not NAD precursors, not glycine plus NAC — has documented all-cause mortality benefit in a powered RCT in a healthy aging population. Healthspan and biomarker improvements have been reported across multiple compounds. Lifespan extension has not.
• How much of "exceptional longevity" is intrinsic biology vs. age-reporting error. A growing body of demographic work suggests that a meaningful fraction of claimed supercentenarians in some regions reflects missing or incorrect birth records rather than true biological exceptionalism. This is unresolved for several of the popular Blue Zone regions and affects how the underlying causal questions can be studied.
• Whether the foundational tier has a ceiling. If you do sleep, food, training, stress, and relationships excellently for fifty years, how high can your healthspan and lifespan go without ever moving into the preservation or experimental tiers? The honest answer is we don't know — almost no one runs that protocol consistently enough for the question to be tested.

The frame that holds up, after all of this, is small: the goal is not to win the longevity lottery. The goal is to stay out of the surgical and replacement bucket as long as possible while the underlying tissue is still responsive. Genes set the slope. The foundational tier moves you along it. The preservation tier extends the runway if the foundation is in place and the supervision is real. The experimental tier is what comes next if the science holds up. None of it is enhancement. All of it is preservation, done early enough to matter.