The premise behind biological age testing is straightforward and appealing. Two people born on the same day can be in wildly different physical states at 60. If some measurable signature captures who is actually aging faster, we could track it, target interventions at it, and see whether we are moving the needle. Whole companies now sell these tests directly to consumers, complete with dashboards and personalized recommendations.
The underlying science is real and moving fast. The consumer product often runs ahead of what the science actually supports.
What the tests are measuring
Several distinct approaches share the biological age label.
- DNA methylation clocks: the most established category, based on the observation that patterns of methylation at hundreds of CpG sites across the genome shift predictably with age. First-generation clocks like Horvath and Hannum were trained to predict chronological age. Second-generation clocks like PhenoAge and GrimAge were trained on health outcomes and predict mortality and disease risk better than chronological age alone. Newer efforts like DunedinPACE attempt to measure the pace of aging rather than a single age estimate.
- Proteomic clocks: use panels of blood proteins to estimate biological age. Some large studies suggest these track health outcomes at least as well as methylation clocks.
- Composite clinical clocks: assemble common lab values (glucose, CRP, albumin, creatinine, and so on) into a single age estimate. The best known is PhenoAge, which was developed from NHANES data.
- Telomere length: once heavily marketed, now considered a relatively noisy and weak predictor at the individual level, though useful at the population level.
Each captures something real. None captures the whole picture, and different clocks can give quite different answers for the same person.
Where they are useful and where they mislead
At the population level, second-generation methylation clocks are the strongest predictors we currently have of morbidity and mortality independent of chronological age. They also move in response to some interventions in trials, most notably CALERIE, where mild caloric restriction slowed the DunedinPACE measure over two years.
The clocks are useful research instruments. Whether they are useful personal tools is a much narrower question.
The individual-level story is messier. Test-retest variability is meaningful. The same sample analyzed twice, or samples taken a few weeks apart, can produce ages that differ by several years without any biological change. Different clocks disagree with each other. And the response to a specific intervention in a specific person is currently not well characterized, meaning the change from a lifestyle experiment may be indistinguishable from noise.
Most consumer products do not surface these uncertainties. They deliver a number, a peer comparison, and a personalized recommendation package. That framing overstates what the underlying measurement can currently support.
What matters if you take a test
If you are curious enough to try one, treat the result like a weather forecast rather than an odometer reading. A few practical points:
- Prefer second-generation clocks (PhenoAge, GrimAge, DunedinPACE) over first-generation ones for anything health-related
- Interpret a single result with substantial skepticism; trends across multiple measurements are more meaningful than any one number
- Do not chase small changes with expensive supplements or protocols; the signal-to-noise ratio does not currently support that
- Recognize that lifestyle factors known to improve health, from sleep to exercise to smoking cessation, generally move these clocks in the expected direction
The larger scientific question
A deeper question hangs over the whole enterprise. Even if a clock reliably predicts mortality, that does not automatically mean slowing the clock will extend life. The clock could be a downstream marker of an underlying process without being causal. Interventions that move the marker without addressing the underlying biology would look effective without being so.
This is not hypothetical. Several biomarkers in medical history have looked like plausible surrogate endpoints, only for interventions that moved them to fail to improve hard outcomes. Whether biological age clocks are causal proxies or merely correlated ones is one of the central open questions in the field.
The bottom line
Biological age clocks are useful research tools and interesting personal experiments, but consumer marketing runs well ahead of the evidence. Trends matter more than single numbers, second-generation clocks outperform first-generation ones, and lifestyle basics still do more than any test can measure.