Your Genes Matter More Than You Were Told

A new paper in Science just overturned one of the most cited claims in longevity

For years, a single statistic has dominated the longevity conversation. You’ve heard it from podcasters, bestselling authors, Netflix docuseries, and even some researchers who should have looked more carefully at the data. It goes something like this.

“Only about 20 percent of how long you live is determined by your genes. The other 80 percent is lifestyle.”

Or worse, the version that emerged after a 2018 paper from Calico Life Sciences and Ancestry.com.

“Genes account for less than 7 percent of your lifespan.”

These claims were repeated so often, and with such confidence, that they became foundational assumptions in both pop science and serious longevity research. I cannot tell you how many times I sat in meetings, read articles, or listened to interviews where someone cited the 7 percent figure as proof that studying the genetics of aging was essentially a dead end. That genetics was not where the action was. That lifestyle explained almost everything.

As someone who has spent her career studying the genetics of brain aging, this narrative was maddening. Not because lifestyle doesn’t matter. It does. But because lifespan is an inherently noisy variable, and the people amplifying these low estimates rarely stopped to consider what that noise was doing to the measurement.

A new paper published in Science in January 2026 has finally set the record straight.

What the new study found

The study, led by Ben Shenhar from Uri Alon’s lab at the Weizmann Institute of Science, used mathematical modeling combined with data from three major twin cohorts in Denmark, Sweden, and the United States. The researchers found that the heritability of intrinsic human lifespan is approximately 50 to 55 percent. That is more than double the previous consensus estimate and nearly eight times larger than Calico’s widely cited figure.

The key insight was deceptively simple. Previous studies, including the twin studies and the Calico pedigree analysis, did not separate deaths caused by aging biology from deaths caused by external factors like accidents, infections, violence, and environmental hazards. When someone gets hit by a bus or dies of cholera, that death has nothing to do with their genetic predisposition to aging. But it still shows up in the data and drags down the measured correlation between genetically related individuals.

The Shenhar team used a mathematical framework based on the Gompertz-Makeham mortality model to partition extrinsic from intrinsic mortality. The Gompertz-Makeham model describes overall mortality as the sum of two components. The first is the Makeham term, a constant, age-independent baseline risk of death from external causes like accidents and infections. The second is the Gompertz function, which captures the exponential rise in death probability with age, the mathematical signature of biological senescence. Because the historical twin datasets did not include cause-of-death information, the team could not simply label each death as intrinsic or extrinsic. Instead, they exploited a known feature of human mortality curves. Between roughly ages 20 and 40, overall mortality hits a plateau driven primarily by extrinsic causes, before the exponential aging curve takes over. The value of that plateau allowed them to estimate the extrinsic mortality rate and mathematically subtract its effects.

The researchers applied this framework to data from three twin cohorts. The Danish Twin Registry, with twins born between 1870 and 1900. The Swedish Twin Registry, with twins born between 1886 and 1925. And the Swedish Adoption/Twin Study on Aging (SATSA), which critically included twins raised apart. This last dataset was the key validation step. Identical twins raised in different households share their genetics but not their environment, making them a natural experiment for disentangling the two. Across all three cohorts, once extrinsic mortality was accounted for, the heritability of lifespan consistently converged around 50 percent. The researchers also showed that heritability estimates increased in more recent birth cohorts, as expected given the historical decline in extrinsic mortality.

The problems with the Calico paper

The 2018 Calico paper by Ruby et al. analyzed pedigree data from over 400 million individuals in Ancestry.com’s public family trees. It was an impressive dataset in terms of sheer scale. The headline finding was that lifespan heritability was “well below 10 percent” after correcting for assortative mating, the tendency of people to choose partners with similar traits.

The paper had several significant limitations that the new Shenhar study highlights.

The data was from the 1800s and early 1900s. To be clear, Calico and Ancestry did not analyze anyone’s DNA. They used family tree data, birth years, death years, and familial connections, compiled by Ancestry’s users from historical records. Heritability was inferred from lifespan correlations among relatives within those pedigrees. The problem is that the individuals in these trees were overwhelmingly from birth cohorts in the 1800s and early 1900s, when extrinsic mortality was dramatically higher than it is today. Infectious disease, poor sanitation, occupational hazards, and violence claimed lives at rates roughly 10 times higher than in modern populations.

No separation of intrinsic and extrinsic mortality. This is the core issue and I will spend more time on it below. The Calico study treated all deaths equally. A death at age 25 from typhoid and a death at age 85 from heart failure were given the same statistical weight.

User-generated genealogy data. The Ancestry trees were created by amateur genealogists, not researchers. The study assumed biological parentage, but as the authors themselves acknowledged, the “social concept of a nuclear family was not rigidly defined” and the data likely included adoptive relationships, step-families, and errors in lineage recording.

Assortative mating correction may have overcorrected. The paper’s main methodological contribution was showing that in-laws had similar lifespan correlations to blood relatives, which they attributed to assortative mating. But the Shenhar paper demonstrates that this pattern is also expected when extrinsic mortality is high, because random external deaths reduce correlations among genetic relatives to the point where they look similar to correlations among non-genetic relatives. The assortative mating effect was real, but it was layered on top of an even larger confound that was not addressed.

Lifespan is a dirty variable

This is the point that should have been obvious all along, and the reason these low estimates drove me crazy for years.

Lifespan is a dirty variable. It is the amalgamation of death by a huge number of different causes. Car accidents. Infections. Cancer. Heart disease. Neurodegeneration. Workplace injuries. War. Each of these has its own risk profile, its own genetic and environmental architecture. When you lump them all into a single number, the year someone died, you are introducing massive noise into your measurement. And noise suppresses measured heritability. That is basic statistics.

We already knew that individual diseases of aging are substantially heritable. Alzheimer’s, cardiovascular disease, many cancers, type 2 diabetes. These all have strong genetic components. It was always paradoxical that the diseases people die from would be highly heritable, but the act of dying from them somehow would not be. That paradox should have been a clue that the lifespan variable itself was the problem, not the genetics.

The low heritability estimates did not mean that genetics has little influence on aging. They meant that lifespan, as measured in those studies, was too noisy a variable to detect the genetic signal. There is a world of difference between “genes don’t matter for how long you live” and “we measured a messy variable badly and the genetic contribution looked small.” The field chose the first interpretation and ran with it.

How this shaped the longevity narrative

Dan Buettner, creator of the Blue Zones brand, built an empire on the claim that “only about 20 percent of how long the average person lives is dictated by genes.” He cited the Danish twin studies as his source. After the Calico paper came out, some versions of this talking point dropped even further. In at least one interview, Buettner’s team claimed that “daily habits rather than our genes actually account for as much as 90 percent of our life expectancy.”

Peter Attia, in his bestselling book Outlive, framed genetics as playing a relatively modest role in longevity, consistent with the prevailing estimates at the time.

The narrative became self-reinforcing. If genetics only contributed 7 to 20 percent, then why fund expensive genetic studies of aging? Why develop genetic risk assessments? Why even bother looking at someone’s DNA when you could just tell them to eat more vegetables and walk 10,000 steps? I even had to field this critique during my pitch to be admitted to Berkeley SkyDeck’s accelerator program. The idea that lifespan heritability was negligible had become so entrenched that I was being asked to justify the scientific basis of genetic risk assessment for brain aging in a startup pitch. As Shenhar himself noted, “low heritability estimates may have discouraged funding and research into the genetics of aging, suggesting it was largely random or environmental.”

What 50 percent heritability actually means

It is important to be precise about what this number represents. Heritability of 50 percent does not mean that 50 percent of your lifespan is “determined” by your genes in some fixed, fatalistic sense. It means that approximately half of the variation in lifespan across a population can be attributed to genetic differences between individuals.

This is actually a very normal number for a complex human trait. Height is about 80 percent heritable. BMI is around 40 to 70 percent. Personality traits like introversion hover around 50 percent. The age of menopause onset, which is an aging-related physiological change, is also about 50 percent heritable. What was always strange about the old estimates was that lifespan seemed to be a bizarre outlier, with much lower heritability than everything else. The Shenhar paper resolves that inconsistency. As Shenhar put it, they are “bringing back life span, which was thought to be very different, into the same playing field with the rest of the traits.”

This also aligns with what we see in animal models. In laboratory mice and flies, where you can tightly control the environment, lifespan heritability is consistently high. The disconnect between animal data and human estimates was another red flag that something was off with the human studies.

As co-author Joris Deelen emphasized, environment still matters. Having a genetic propensity for longevity does not guarantee a long life. But it does mean that the genetic signal is strong enough to be worth studying, and strong enough to inform personalized medicine. As the accompanying Science commentary by Bakula and Scheibye-Knudsen stated, a substantial genetic contribution “strengthens the rationale for large-scale efforts to identify longevity-associated variants, refine polygenic risk scores, and link genetic differences to specific biological pathways that regulate aging.”

The bottom line

Your genes matter for how long you live. A lot more than you were told. The old estimates were not wrong in the sense that they accurately measured what they measured. But they were measuring a dirty variable and interpreting the noise as evidence that genetics was unimportant. That was always a flawed inference, and now we have the data to prove it.

This paper does not diminish the importance of lifestyle. Exercise, sleep, diet, social connection, stress management, and cognitive engagement all matter for brain health and longevity. I practice all of these things rigorously in my own life. But pretending that genetics is a minor footnote in the longevity story was always inconsistent with what we knew about centenarian families, animal models, and virtually every other complex human trait.

If you have been told not to worry about your genetics because “lifestyle is 80 percent of longevity,” I would encourage you to revisit that assumption. Understanding your specific genetic risk factors is not fatalistic. It is the first step toward doing something about them.