What is biological age?

In one long-running New Zealand study, 964 adults were assessed at around age 38. Researchers compared eleven different measures of biological aging. People of the same age differed on every one of them.^[1]

Why would people born in the same year differ? Aging is the gradual build-up of changes in cells and tissues, and it doesn't happen at the same pace or in the same way in everyone.^[2] Illness, surroundings and things outside a person's control can all shape a result, not only their choices.^[3]

The term biological age tries to put a number on that difference. Your chronological age is the years since you were born. Your biological age is an estimate of how far along your body seems to be in aging. It is not a thing in the body that a test reads out, like height or blood sugar. A statistical model produces it from measurements of your blood, cells or body.^[3][4]

That one fact explains most of the confusion around these tests. Once you know what a model was taught to estimate, the number on a report starts to make sense.

Why do researchers want a biological age?

Researchers who want to know whether a treatment slows aging itself, rather than one disease at a time, have a practical problem. A trial that waits for disease or death can take decades.^[2] They want a measurement that could show a change much sooner. Biological age is the name for that goal, and whether any measure has reached it is still being tested.^[5]

Can a test measure your biological age?

Not directly. A test measures something: chemical tags on your DNA (epigenetic clocks), proteins in your blood (proteomic clocks), routine blood-test values (phenotypic clocks) or sugar patterns on antibodies (glycomic clocks). A model then turns those measurements into a number.

A useful measure should at least tell you something beyond your calendar age.^[4] There is no agreed definition of biological age and no reference method yet.^[3][4]

What does the number estimate?

A model is only as meaningful as what it was taught to predict. Most results fall into three kinds, and the kind changes what the number means. Researchers picture two sorts of measure: an odometer, which shows how much aging has built up, and a speedometer, which shows how fast it is happening. These are analogies for what a model aims to estimate, not readings of damage or speed.^[2] Age-like and risk-like results are odometer-like; a pace result is speedometer-like.

Age-like. Early clocks, such as Horvath's, were taught to predict calendar age from DNA methylation: chemical tags on DNA that can influence gene activity.^[6] A clock that predicted calendar age perfectly would carry no information about biological aging beyond the calendar age itself.^[4]

Risk-like. Other clocks were taught to predict health outcomes. GrimAge combines calendar age, sex and methylation-based estimates of smoking and of blood proteins into a score related to the risk of dying, then rescales that score into years.^[7] DNAm PhenoAge was trained to predict a similar risk-related score built from calendar age and nine routine blood measures.^[8] The result is given in years, but a higher number means a higher estimated risk. It does not mean you have that many years left.

Pace. DunedinPACE estimates how fast aging is happening from a single blood sample. It was fitted to a pace measured over 20 years in one New Zealand group.^[9] A pace of 1.0 is the average of the group the clock was built on, so 1.1 would be about 10% faster than that average. It is an estimate, not an exact personal speed.

A report's fine print or the provider's documentation should name the clock behind a number, and the name tells you which kind you have. No kind is better in absolute terms: the question you want answered decides which is more useful. Why aging clocks disagree explains the three in more depth. To go deeper on the inputs, see how epigenetic clocks work, biological age from routine blood work and digital biomarkers of ageing.

What does the gap from your age mean?

Results are often shown as a gap: a biological age of 47 for someone who is 50. Researchers often use an age-adjusted version instead, the difference between a person's score and the score expected for their age, called a residual. Either can be above or below zero.^[4][7] Researchers call this difference “age acceleration”. Despite the name, it describes a difference at one moment, not a speed.^[2][4]

One more thing shapes the gap. Age-trained clocks tend to read older people as younger than they are and younger people as older. So researchers account for calendar age when they test whether gaps relate to health.^[10]

Is there one biological age or several?

Not clearly. In the New Zealand study the eleven measures agreed only weakly, and the authors concluded that the variation did not appear to reflect a single aging process.^[1] In a Swedish twin study that followed people for 20 years, nine biological ages were mostly only weakly related once calendar age was taken out.^[11] A study of 15 clocks in about 1,000 people from Orkney used different kinds of biological data (called omics). It concluded that biological age is measurable and predicts later disease, while allowing that a person may have more than one.^[12] Together, these results show that the scores carry information about later health. They do not settle whether there is one biological age underneath.

Not every “age” on a report comes from a molecular clock. A heart age re-expresses cardiovascular risk over 5 or 10 years as the age of someone with the same estimated risk and a specified set of risk factors.^[13] It summarizes something different from an age-trained DNA clock, so a lower result on one does not guarantee a lower result on the other. A frailty index is the share of assessed health problems a person has, and researchers can convert it into an age using a chosen reference group.^[14]

Even inside one person, estimated ages can differ. One research group's blood-protein models gave separate ages for 11 organs, and the age gaps of different organs were only loosely related. Those were research models; the study does not validate any particular consumer test.^[15] More in organ ages.

What can a result tell you?

In large studies, clocks built on health measures, such as GrimAge and PhenoAge, are linked to later disease and death even after taking calendar age into account. Clocks built only to predict age show weaker, less consistent links.^[11][16] The link is modest. In a representative sample of older Americans, a refined version of GrimAge accounted for less than 5% of the differences between people in each health outcome studied. For most outcomes, other measured characteristics told more.^[16]

These are findings about groups. A 2025 perspective argues that this makes one person's result hard to interpret.^[3] A single result also carries noise. A 2026 analysis of studies of short-term exposures such as meals, stress and altitude found that DNA methylation clock results varied between samples taken hours or days apart. They varied more than when one sample was run twice.^[17] And some clocks were built on groups that may not resemble you.^[3]

More on reading a result: what your test results actually mean, the limitations of aging clocks and how often to test.

Can biological age change?

Scores can move a little. In CALERIE, a two-year trial in healthy adults without obesity, cutting calories slowed DunedinPACE slightly on average, while PhenoAge and GrimAge did not change significantly.^[18] DO-HEALTH was a three-year trial in adults aged 70 and over. In its Swiss substudy, home strength training did not significantly improve the clock results compared with a flexibility program. Omega-3 supplements were linked to a small slowing on some clocks; the researchers say what this means for long-term health is unknown.^[19]

A lower score alone has not been established as proof that an intervention improves health or lifespan.^[2][20] See what the evidence says about exercise and other interventions.

Common questions

Why do two tests give me different numbers?

Two tests can give different numbers because they were built to predict different things (an age, a risk or a pace) from different data (DNA marks, blood proteins, routine blood tests). A difference doesn't by itself mean either test is wrong. Differences in calibration can also play a part. Why aging clocks disagree

Which test should I choose?

It depends on what you want to know: an age-like, a risk-like or a pace result. Our short quiz suggests a type of test to start with, and Compare shows what each test measures.

How often should I retest?

A small change between two tests may be noise. How often to retest depends on why you are testing. How often to test

Is biological age a diagnosis?

No. It is a statistical estimate, and a result shouldn't be used on its own to make medical decisions.

The bottom line

Biological age is an idea with real science behind it, and it can show differences between people of the same age that calendar age misses. But what the number means depends on what the model behind it was taught to predict.

So when you see a biological age, first find out which kind of number it is. Then read it as a rough estimate drawn from groups of people, not as a verdict on your body. That is the first step toward making sense of the growing market of tests and claims built around it.

Sources

Peer-reviewed sources for the scientific statements above, with the commercial ties their authors declare.

1. Belsky et al. (2018). Eleven telomere, epigenetic clock, and biomarker-composite quantifications of biological aging: do they measure the same thing? American Journal of Epidemiology. One New Zealand cohort of 964 adults assessed around age 38; the authors report no competing interests.
2. Moqri et al. (2023). Biomarkers of aging for the identification and evaluation of longevity interventions. Cell. Consensus review by the Biomarkers of Aging Consortium; several authors declare patents, licences or company roles.
3. Apsley et al. (2025). From population science to the clinic? Limits of epigenetic clocks as personal biomarkers. Epigenomics. A perspective article with no new data; the authors report no relevant financial involvement.
4. Sluiskes et al. (2024). Clarifying the biological and statistical assumptions of cross-sectional biological age predictors: an elaborate illustration using synthetic and real data. BMC Medical Research Methodology. Statistical analysis and simulation by an academic group; no competing interests declared.
5. Moqri et al. (2024). Validation of biomarkers of aging. Nature Medicine. Many authors declare ties to clock developers or test companies.
6. Horvath (2013). DNA methylation age of human tissues and cell types. Genome Biology. The author is a named inventor on a university patent application.
7. Lu et al. (2019). DNA methylation GrimAge strongly predicts lifespan and healthspan. Aging. Two authors are named inventors on a university patent application.
8. Levine et al. (2018). An epigenetic biomarker of aging for lifespan and healthspan. Aging. Two authors are named inventors on a university patent application.
9. Belsky et al. (2022). DunedinPACE, a DNA methylation biomarker of the pace of aging. eLife. Several authors are named inventors of DunedinPACE, which the paper says is licensed to a commercial company.
10. El Khoury et al. (2019). Systematic underestimation of the epigenetic clock and age acceleration in older subjects. Genome Biology. No competing interests declared.
11. Li et al. (2020). Longitudinal trajectories, correlations and mortality associations of nine biological ages across 20-years follow-up. eLife. An academic study of a Swedish twin cohort.
12. Macdonald-Dunlop et al. (2022). A catalogue of omics biological ageing clocks reveals substantial commonality and associations with disease risk. Aging. One author declares paid consultancy for two companies.
13. Bonner et al. (2021). Interventions using heart age for cardiovascular disease risk communication: systematic review of psychological, behavioral, and clinical effects. JMIR Cardio. The first author wrote some of the included studies; a co-author works for an organization that built a heart-age tool.
14. Mitnitski et al. (2002). Frailty, fitness and late-life mortality in relation to chronological and biological age. BMC Geriatrics. No competing interests declared.
15. Oh et al. (2023). Organ aging signatures in the plasma proteome track health and disease. Nature. A university patent application was reported; several authors co-founded Teal Omics, with equity, and some declare other company ties.
16. Faul et al. (2023). Epigenetic-based age acceleration in a representative sample of older Americans: associations with aging-related morbidity and mortality. PNAS. No competing interests declared.
17. Sehgal et al. (2026). Biological Versus Technical Reliability of Epigenetic Clocks and Implications for Disease Prognosis and Intervention Response. Aging Cell. Two authors are co-inventors of another clock and have consulted for test companies.
18. Waziry et al. (2023). Effect of long-term caloric restriction on DNA methylation measures of biological aging in healthy adults from the CALERIE trial. Nature Aging. Two authors are named DunedinPACE inventors; the declaration also reports commercial licensing.
19. Bischoff-Ferrari et al. (2025). Individual and additive effects of vitamin D, omega-3 and exercise on DNA methylation clocks of biological aging in older adults from the DO-HEALTH trial. Nature Aging. One co-author (Horvath) founded the foundation that licenses clock patents, including GrimAge, and works for Altos Labs; the other authors declare no competing interests. Co-authors also include the developers of GrimAge and DunedinPACE.
20. Cummings and Kritchevsky (2022). Endpoints for geroscience clinical trials: health outcomes, biomarkers, and biologic age. GeroScience. One author has consulted for a biotechnology company.

Sources checked 2026-11-10. See how we make and check content or report a correction.