Biological Age Tests: What They Measure and How to Choose

You open the report and see a large “52 years”. You have just turned 44. For a few seconds, that number feels more real than your date of birth. Have you suddenly aged by eight years? No.

A biological age test does not uncover a secret age inside your body. It applies an algorithm to a sample or set of measurements and compares them with reference data. The result may be interesting, but it changes with what is measured, how it is processed and what the model was designed to predict.

This guide takes away the drama without taking away the science. We will unpack the main types of test, explain why two reports can disagree and give you a practical checklist before you pay or change anything because of a score.

Chronological and biological age answer different questions

Chronological age counts time since birth. It is boringly precise. “Biological age” tries to summarize differences in processes, function or risk that appear among people with the same number of years.

That does not create a second date of birth. A biomarker of aging is a measurement, or combination of measurements, being evaluated as a representation of some part of the aging process. The Biomarkers of Aging Consortium framework emphasizes that a lack of standards and consensus still holds back the use of many biomarkers in trials and clinical practice.

“What is my biological age?” is therefore too broad. Try this instead: what does this product calculate, in whom was it validated and what decision can I make with reasonable confidence?

Four results that often share one label

The box may say “biological age test”, even though the number belongs to a very different category. They should not be compared as if they were the same unit.

Result type	The model's question	A cautious reading
Age-prediction clock	What chronological age do these signals predict?	A difference from the calendar, not a complete measure of health
Risk-associated clock	How does this pattern relate to death or disease in cohorts?	A statistical association, not your personal prognosis
Pace clock	What rate of change does this pattern reflect?	An estimate of pace according to that specific model
Clinical or functional measure	How are blood pressure, glucose, strength or aerobic capacity?	Data that are often easier to confirm and turn into action

Blood pressure or a strength test does not become more useful simply because it is translated into “years”. It is often better to keep the original measurement, compare it with suitable references and follow its trend. The same applies to aerobic capacity and VO2 max: a specific function provides different information from a molecular algorithm.

What Horvath, PhenoAge, GrimAge and DunedinPACE measure

Epigenetic clocks use DNA methylation patterns. They are algorithms trained to predict a target, not sensors that directly read total wear across the body.

• Horvath: it was trained across multiple tissues to estimate chronological age from 353 DNA positions. It was a major methodological advance, but the original paper already noted that the biological meaning of the signal was not yet clear.
• PhenoAge: it began with a “phenotypic age” built from age and clinical biomarkers related to mortality, then created a methylation predictor of that construct.
• GrimAge: it combines epigenetic surrogates for proteins and smoking exposure to predict time-to-event outcomes in cohorts. An association with mortality does not mean it can calculate when one person will die.
• DunedinPACE: it was designed from longitudinal change in 19 organ-system indicators in a cohort and distilled into a blood methylation measure. Its output aims to represent pace, not accumulated years. Our guide to DunedinPACE explains how to read values such as 0.9 or 1.1 without treating them as a literal speedometer.

All four names belong to the same field, but they optimize different outcomes. Choosing whichever gives you the youngest number is not choosing the most valid test. It is changing the question after seeing the answer.

Do they work for one person or only for research groups?

A model can correctly separate groups with higher and lower risk and still be too inaccurate to guide one person. That is the difference between population validity and individual utility.

A 2025 critical review concluded that epigenetic clocks have been valuable for population science, while technical and biological properties currently limit their use in individual decisions. The concerns include sample collection and processing, algorithm implementation, tissue, developmental stage and environmental or sociodemographic context.

Repeatability matters too. In a study of six clocks, technical noise produced deviations of up to nine years between replicates. The authors developed principal-component versions that greatly reduced this variation. The improvement is promising, but it also shows why a small difference should not be sold as rejuvenation.

This does not invalidate the research. It puts each tool in its proper place: excellent for generating knowledge, potentially useful as an exploratory result and still insufficient on its own to diagnose accelerated aging or prescribe a treatment.

Why two tests can give you different ages

Imagine that one uses blood and another saliva. One tries to resemble your chronological age; another incorporates smoking and protein signals; a third calculates pace. Even if all three print “years”, they are not weighing the same object.

Differences may come from:

• the outcome used to train the algorithm;
• the population, ages and health of the training sample;
• the tissue and mixture of cells in the specimen;
• collection, transport, laboratory and platform;
• data preprocessing and software version;
• technical variation and short-term biological change.

A comparison only begins to be reasonable when method, laboratory, tissue and version remain the same. Even then, you need the expected error. If a provider does not publish repeatability or uncertainty, the decimal in the report creates a level of precision the test may not possess.

A practical checklist before you buy

You do not need to become a biostatistician. These questions separate an explainable product from a beautiful black box:

1. What outcome does it calculate? Chronological age, risk, pace and a proprietary blend are not equivalent.
2. In whom was it validated? Look for sample size, age, sex, background, health and external validation, not only the training set.
3. Which specimen and version does it use? Blood, saliva and cheek cells may produce different readings. The algorithm can change too.
4. How repeatable is it? Ask for replicate variation, an uncertainty interval and the smallest interpretable change.
5. What evidence exists at individual level? A correlation with mortality in a cohort does not establish personal clinical utility.
6. Which decision will it change? Define the action before seeing the result. This helps you avoid shopping for a supplement or treatment to fix a number.
7. What happens to your data? Ask where it is stored, who may use it, whether it is shared and how to request deletion.
8. Does the business depend on retesting or selling you something? That does not invalidate the test, but it deserves a clear disclosure.

How to read your report without letting the number take over

Start with the exact algorithm name and its target. Then find the uncertainty. “Eight years older” means little if expected variation is wide or the product does not report it.

Compare the result with information that already has clinical meaning: symptoms, history, blood pressure, smoking, sleep, activity, medication and tests indicated for your context. If a modifiable risk appears, confirm it with the appropriate method. A clock neither replaces an assessment nor decides for you.

Protect the emotional side too. A high score is not a verdict; a low score is not permission to ignore health signals. The report may open a conversation, but it does not deserve to become an identity.

Can biological age be “reversed”?

Exercise, sleep, nutrition, stopping smoking and treating risk factors can improve health and many measurements. That is a worthwhile goal even if no clock moves.

Some intervention studies report shifts in selected epigenetic scores. A change in an algorithm, however, does not show that the whole body has rejuvenated, that the effect will last or that disease and mortality will fall. Demonstrating benefit requires clinical outcomes, suitable controls, replication and time.

The healthiest rule is simple: do not treat the test, treat the person and confirmed risks. Be wary when someone detects “accelerated age” and sells the exact package to lower it in the same conversation.

When to repeat a test

There is no universal schedule of three, six or twelve months. Timing depends on the question, technical stability, the size of change the test can detect and whether the result will alter a decision.

If you repeat it, use the same specimen type, laboratory, platform, algorithm and version. Keep conditions comparable where possible and record changes in medication or health. Before paying, ask what difference exceeds the noise. If nobody can answer, it may be more sensible to track the function or risk factor you are actually trying to improve.

Frequently asked questions about biological age tests

What is biological age?

It is an estimate built by a model: it may summarize methylation patterns, clinical data, risk or pace of change. There is no single biological age hidden inside the body, and the result is neither a diagnosis nor an individual prediction of lifespan.

How is it different from chronological age?

Chronological age counts time since birth. A biological age test compares selected measurements with a model or reference population. Two people of the same age can differ in risk and function, but that does not make a test score their true age.

What is the best biological age test?

There is no universal winner. The best option depends on the question: predicting chronological age, summarizing risk, studying pace of aging or measuring a specific capacity. Before buying, check independent validation, tissue, version, repeatability, uncertainty and which decision the result will change.

Are epigenetic clocks reliable?

They are valuable for studying groups, but their usefulness for decisions about one person remains limited. Sampling, processing and the algorithm add variation. One study found deviations of up to nine years between replicates across six clocks, although later methods improved repeatability.

Why do two tests give different ages?

They may target different outcomes, tissues and signals, use different training populations or run different algorithm versions. Blood, saliva and cheek cells are not interchangeable. Disagreement does not show that one has found your true age and the other is wrong.

Can I lower or reverse my biological age?

You can improve blood pressure, fitness, sleep, smoking, glucose and other factors that matter to health. Some interventions also move certain clocks, but lowering a score does not prove whole-body rejuvenation or guarantee less disease or a longer life.

What should I do if my result is high?

Do not buy a treatment on impulse. Check what the test calculated, its uncertainty and whether the difference exceeds its expected variation. Then compare it with symptoms, medical history and validated measurements. Address modifiable risks that can actually be confirmed first.

When should I repeat the test?

There is no universal interval. Repeat it only when a change larger than the noise can answer a question or alter a decision. Use the same laboratory, specimen, algorithm and comparable conditions; first ask about repeatability and the smallest change the provider considers interpretable.

Sources

1. Moqri M et al. Biomarkers of aging for the identification and evaluation of longevity interventions. Cell. 2023.
2. Apsley AT et al. Limits of epigenetic clocks as personal biomarkers. Journal of Gerontology: Biological Sciences. 2025.
3. Higgins-Chen AT et al. Epigenetic clock reliability and longitudinal tracking. Nature Aging. 2022.
4. Horvath S. DNA methylation age of human tissues and cell types. Genome Biology. 2013.
5. Levine ME et al. PhenoAge, an epigenetic biomarker for lifespan and healthspan. Aging. 2018.
6. Lu AT et al. GrimAge and its associations with lifespan and healthspan. Aging. 2019.
7. Belsky DW et al. DunedinPACE, a DNA methylation biomarker of the pace of aging. eLife. 2022.

Method: narrative review of the linked validation framework, individual-utility review and original studies. Cohort associations are not presented as personal prognosis, and a score change is not equated with rejuvenation. Sources checked 29 August 2026. This article is educational and does not replace medical assessment.

Commercial disclosure: Progevita may include biomarkers and functional tests in a medical assessment. That does not mean everyone needs a biological age test or that a score justifies purchasing a treatment.