Biological age: what the tests really measure, and what they don’t tell you

Your chronological age is on your passport. Biological age starts from a simple observation: two people born the same year do not age at the same pace. The US National Institute on Aging describes it as how old our cells, tissues and organ systems appear to be, based on their biochemistry.

Yet this age cannot be read off anywhere. It is calculated: a statistical model, developed on thousands of people, turns biological measurements into a number of years. The result depends on the model, the data it was built on and what it was trained to predict. “Biological age” is not one measurement but a family of estimates, and they do not all answer the same question.

Measures based on a blood test

The most studied is Phenotypic Age, or PhenoAge. A 2018 study in PLOS Medicine calculates it from chronological age and nine routine blood markers:

• albumin, creatinine and glucose;
• C-reactive protein (CRP), a marker of inflammation;
• lymphocyte percentage and white blood cell count;
• mean cell volume and red cell distribution width;
• alkaline phosphatase.

The result is expressed in years: the age that, in the reference population, matches the person’s estimated mortality risk. Among more than 11,000 US adults followed for about twelve years, each additional year of Phenotypic Age, at the same chronological age, was associated with a 9% higher risk of death from any cause. The association held in seemingly healthy participants, with no reported disease and a normal BMI. The authors point to the main limitation themselves: no repeated measurements over time, and the need to confirm the findings in other cohorts.

Each of these markers also means something on its own, and is read in the context of the sample: that is what our guide on how to read blood test results covers.

Epigenetic clocks

The second family reads DNA methylation: chemical tags on DNA that shift with age without changing the genetic code. In 2013, Steve Horvath built a clock from 8,000 samples, based on 353 sites in the DNA, that estimates the “DNA methylation age” of most tissues.

Later clocks were trained to predict something other than chronological age. GrimAge, published in 2019, combines DNA-based estimates of seven blood proteins and of smoking history; it is associated with time to death, coronary heart disease and cancer. DunedinPACE, published in 2022, gives a pace rather than an age. It comes from a New Zealand cohort followed over two decades on 19 indicators of organ function. A value of 1 means one year of biological aging per calendar year; above 1, aging is faster.

These clocks are also offered by companies that sell tests directly to consumers. According to a 2025 analysis in Epigenomics, such tests often use saliva or a cheek swab, which are easier to collect, whereas most clocks were trained on blood, whose methylation profile is different.

A biological age test is first of all a lab measurement, with its share of noise. A 2022 study in Nature Aging measured the same sample twice: for six widely used epigenetic clocks, technical noise alone produced gaps of up to 9 years between the two results. The authors propose a way of computing the clocks that brings most of these gaps under 1.5 years. DunedinPACE, for its part, was built by leaving out the least reproducible DNA sites. Reliability therefore depends on the clock and on how it is computed.

The Epigenomics analysis goes further. Because they are trained on different aspects of aging, clocks often give conflicting estimates for the same sample. A recent illness, stress or a medical treatment can skew a reading; two labs that process the data differently can get different results; and there is no agreed cut-off at which a gap becomes clinically meaningful. The authors conclude that, as things stand, these clocks should not be used to make decisions about an individual.

The issue goes beyond epigenetics. Researchers from the Biomarkers of Aging Consortium, in a 2024 review in Nature Medicine, note that there is still no consensus on how biomarkers of aging should be validated before they are used in the clinic. With PhenoAge, the lab tests are routine, but the estimate remains a population statistic, validated without repeated measurements.

“Turn back the clock by ten years”: the promise is everywhere. In humans, the evidence is thin. One example is the CALERIE randomized trial, whose analysis published in 2023 in Nature Aging looked at epigenetic clocks. In all, 220 adults without obesity were randomly assigned either to cut their calorie intake by 25% for two years or to keep eating as usual. In practice, the first group reduced its intake by about 12% on average. The authors note that earlier findings came mostly from small studies without a control group.

The result: a pace of aging, measured by DunedinPACE, slowed by 2 to 3%. The other clocks tested, including GrimAge, did not change significantly. The authors describe small effects, in an analysis carried out after the fact, among healthy volunteers who do not represent the general population. Above all, they point out that a conclusive answer will require long trials measuring the onset of chronic disease and mortality.

Moving a marker, in other words, is not proof of a longer life in good health. And the calorie restriction in the trial was prescribed and monitored in a research setting: it is not something to try on your own. The National Institute on Aging urges caution with anti-aging interventions, especially when they seem too good to be true.

A biological age is not a target to hit. Used with some distance, it can open a conversation with a doctor: which markers drive the result, what do they say about your risk factors, what is worth keeping an eye on? More often than not, the answer leads back to levers you already know.

Those levers are the same whatever the number. The National Institute on Aging sums them up plainly: eat well, exercise, get enough sleep, cut back on unhealthy habits and stay socially connected. For fitness, our article on VO2 max and longevity goes through what is known. And if you see a doctor practicing longevity medicine, the same question applies to every test on offer: what evidence is it based on?

Finally, a single measurement says little; its trend, measured the same way, says more. Tracking your data over time to estimate a risk is the idea behind predictive health.

That is the approach behind MyTwin Longevity, an app built on MyTwin, the health digital twin: it estimates biological age with PhenoAge from a blood check-up, and has it reviewed with a doctor specialized in prevention and longevity. The full journey is described in the launch of MyTwin Longevity, told by the MyTwin Lab.