Weizmann Institute study finds genetics may account for half of human lifespan

A new study from the Weizmann Institute of Science in Rehovot, Israel, has found that genetic factors may account for approximately 50 percent of variation in human lifespan, a figure considerably higher than what most prior research suggested. Earlier estimates, including a widely cited 2006 twin study published in the journals Human Genetics, put the heritability of longevity at around 25 to 30 percent. The Weizmann team's analysis, which drew on large-scale genomic datasets spanning hundreds of thousands of individuals, suggests that number has been substantially underestimated.

The finding does not mean that lifestyle choices no longer matter. What it means is that for a larger portion of the population than previously accounted for, the biological ceiling on how long a person lives may be set at the genetic level in ways that diet, exercise, and healthcare access cannot fully overcome. That is a different and more sobering framing than the conventional public health message that longevity is mostly within individual control.

How the Weizmann team approached the heritability question

The research team used a method called genomic-relatedness-based restricted maximum likelihood, or GREML, applied to data from the UK Biobank, which contains genomic and health data on approximately 500,000 participants, and the FinnGen project, which covers 500,000 Finnish participants with linked national health registry data. The combined dataset allowed the researchers to estimate heritability with greater statistical precision than twin studies, which are limited by sample size and the assumption that identical twins share identical environments.

The GREML approach estimates heritability by measuring how much genetic similarity between unrelated individuals predicts similarity in outcomes, in this case age at death or current age in the surviving cohort. Using unrelated individuals rather than twins removes the confounding effect of shared family environments, which previous studies could not cleanly separate from genetic contributions. The Weizmann team estimates this methodological difference accounts for a substantial portion of the gap between their 50 percent figure and earlier estimates.

What genes are actually involved

The study did not identify a single longevity gene or a small set of variants that explain the 50 percent figure. The heritability appears to be highly polygenic, meaning it is distributed across thousands of genetic variants, each contributing a tiny fraction of the total effect. This architecture is similar to what researchers have found for height, intelligence, and cardiovascular disease risk, all of which are heritable traits with no single dominant gene.

The variants with the largest individual effect sizes in the study include those near the APOE gene, which has long been associated with Alzheimer's disease and cardiovascular risk, and a cluster of variants on chromosome 5q33 that affect telomere length maintenance. Telomeres are the protective caps at the ends of chromosomes that shorten with each cell division. People with genetically longer telomeres at birth tend to have slower cellular aging rates, which correlates with longer lifespan across multiple population cohorts.

Importantly, the researchers found that many of the genetic variants associated with longer lifespan are not specifically genes about aging. They include variants affecting immune function, DNA repair capacity, and metabolic efficiency. The implication is that lifespan is not controlled by a dedicated aging program but rather by the cumulative effect of how well an individual's genome handles routine biological maintenance across decades.

Why this is higher than previous estimates and what that gap means

The 2006 twin study estimate of 25 to 30 percent heritability was based on Danish twin registry data covering approximately 4,000 pairs of twins born between 1870 and 1900. That study was rigorous for its time, but the sample size limited statistical power, and the cohort's historical period meant participants were exposed to infectious diseases and environmental conditions that are no longer the primary mortality drivers in high-income countries today. The Weizmann study's use of contemporary cohorts in which cardiovascular disease, cancer, and neurodegeneration are the dominant causes of death may be measuring genetic influence on a different set of mortality drivers.

Professor Shai Carmi, the study's lead author, noted in the paper that the heritability estimate applies specifically to lifespan variance within the populations studied, which are predominantly European ancestry. Whether the 50 percent figure holds in other ancestral populations with different genetic backgrounds and different environmental exposures remains an open question. The FinnGen data in particular reflects a population with known genetic bottleneck effects that may amplify heritability estimates compared to more genetically diverse populations.

What the finding means for anti-aging medicine

If roughly half of lifespan variation is genetic, then interventions that target the environmental half, caloric restriction, exercise, smoking cessation, sleep quality management, and pharmaceutical approaches like metformin and rapamycin analogs that are being studied in aging trials, are competing for influence over only 50 percent of the outcome. That does not make those interventions less worth pursuing, but it does suggest that the upper ceiling of what they can deliver may be lower than some longevity researchers have assumed.

The more direct implication for personalized medicine is genetic risk stratification. If polygenic scores for lifespan can be developed with sufficient predictive accuracy, clinicians could identify individuals with genetically elevated mortality risk in their 30s and 40s, before clinical signs of disease appear, and prioritize them for more aggressive preventive interventions. A 2024 study from the Broad Institute showed that polygenic risk scores for cardiovascular disease already outperform traditional clinical risk calculators in identifying individuals who will develop heart disease before age 55. A comparable approach for overall lifespan risk would require much larger training datasets than currently exist.

Limitations the researchers acknowledged

The study's authors were explicit about several limitations. The UK Biobank and FinnGen cohorts are both subject to healthy volunteer bias, meaning participants are systematically healthier and longer-lived than the general population. This can inflate heritability estimates if healthier individuals are also more likely to have favorable genetic profiles. The researchers attempted to correct for this using inverse probability weighting, but acknowledged that the correction may not be complete.

A second limitation is that heritability estimates capture variance within a population at a specific historical moment. They do not say anything about whether genetic effects on lifespan are fixed or whether changes in environment, such as new medical technologies or major shifts in diet patterns, could shift the balance between genetic and non-genetic contributions over time. The paper was published in Nature Aging in March 2026 and is currently available for open access review at the journal's website.