Biological Age: What Epigenetic Clocks Measure, What They Cannot, and the Markers Operators Should Track Instead
Biological age is a real scientific idea wrapped in a consumer product that cannot yet deliver what the label promises.
Operators love a single number. Revenue, runway, resting heart rate. So when a test arrives in the post promising to tell you that your body is 38 while your passport says 46, it sells itself. The number feels like a verdict on the discipline.
The science underneath those tests is serious. Epigenetic clocks are among the most important tools in ageing research, and the papers that built them are quoted here directly. But the gap between what a clock measures in a study of thousands and what one saliva kit tells one person about one year is wide, and the companies selling the kits do not lead with that.
This article explains what the clocks measure in the words of the people who built them, the difference between the first and second generations, what the reliability research says about test-retest noise, what is known about moving the number, and the cheaper markers with far stronger mortality data that an operator should track first. Education, not diagnosis: any blood or saliva result is interpreted with your doctor.
What an epigenetic clock measures
DNA methylation is a chemical mark on DNA that changes with age in patterns regular enough to predict it. The first multi-tissue clock was published in 2013 in Genome Biology. Its author describes the build: "I developed a multi-tissue predictor of age that allows one to estimate the DNA methylation age of most tissues and cell types. The predictor, which is freely available, was developed using 8,000 samples from 82 Illumina DNA methylation array datasets, encompassing 51 healthy tissues and cell types."
The clock rests on "the 353 CpG sites that together form an aging clock." Its properties, as the paper lists them: "first, it is close to zero for embryonic and induced pluripotent stem cells; second, it correlates with cell passage number; third, it gives rise to a highly heritable measure of age acceleration; and, fourth, it is applicable to chimpanzee tissues."
The interpretation the author offers is careful: "I propose that DNA methylation age measures the cumulative effect of an epigenetic maintenance system." Not a direct measure of how worn out you are. A readout of a maintenance process, trained to predict calendar age.
That last point matters. A clock trained on chronological age is, by construction, trying to guess your birthday. The difference between its guess and your real age, called age acceleration, is the interesting residual. It is also where the noise lives.
First-generation versus second-generation clocks
The 2018 paper that introduced DNAm PhenoAge in Aging names the shift: "While the first generation of epigenetic biomarkers of aging were developed using chronological age as a surrogate for biological age, we hypothesized that incorporation of composite clinical measures of phenotypic age that capture differences in lifespan and healthspan may identify novel CpGs and facilitate the development of a more powerful epigenetic biomarker of aging."
Second-generation clocks are trained not on your birthday but on outcomes: clinical markers, mortality, disease. The result, in the authors' words, was a biomarker that "strongly outperforms previous measures in regards to predictions for a variety of aging outcomes, including all-cause mortality, cancers, healthspan, physical functioning, and Alzheimer's disease."
The 2019 GrimAge paper, also in Aging, went a step further and built the clock from methylation-based estimates of blood proteins and smoking history. "The resulting predictor of lifespan, DNAm GrimAge (in units of years), is a composite biomarker based on the seven DNAm surrogates and a DNAm-based estimator of smoking pack-years." Validated in thousands of people, it "stands out among existing epigenetic clocks in terms of its predictive ability for time-to-death (Cox regression P=2.0E-75), time-to-coronary heart disease (Cox P=6.2E-24), time-to-cancer (P= 1.3E-12)."
Those are population results. Across thousands of people, a higher GrimAge means a shorter expected life. Whether the clock can tell you, one person, what your last year of training did is a different question, and the next section answers it.
What the reliability research says
A 2022 paper in Nature Aging tested what happens when the same sample is measured twice. "Epigenetic clocks are widely used aging biomarkers calculated from DNA methylation data, but this data can be surprisingly unreliable. Here we show technical noise produces deviations up to 9 years between replicates for six prominent epigenetic clocks, limiting their utility."
Nine years, from the same blood, on the same day. The authors proposed a fix, principal-component versions of the clocks, which "show agreement between most replicates within 1.5 years." That is a real improvement and it is not what most consumer kits report.
A 2026 follow-up in Aging Cell separated two kinds of reliability across 18 clocks. Technical reliability, the same sample run twice, was mostly good: "Most clocks demonstrated excellent technical reproducibility across replicate assays." Biological reliability, the same person sampled again days apart under ordinary life, was not: "Biological reliability, assessed across repeated measures collected within short intervals under varying conditions such as meals, stress, and environmental exposures, was substantially lower. Most clocks showed only low to moderate stability which further decreased when adjusting for immune composition."
Two more lines from that paper should be printed on every kit. "Notably, technical reproducibility did not predict biological reliability." And: "Clocks with higher reliability produced more stable associations with cognitive outcomes and more consistent responses to interventions, while less reliable clocks yielded variable or misleading results."
For an operator the translation is simple. If you test in January after a hard quarter and again in July after a good one, part of the difference is you, part is the meal you ate that morning, and part is the assay. The research cannot yet tell you which part is which. One author of both reliability papers discloses commercial ties to a company that sells clock-based tests, which is worth knowing and does not change the finding.
What moves the number
Honest answer: less is established than the marketing implies.
At population scale, the GrimAge paper reports that its biomarkers "show the expected relationship with lifestyle factors including healthy diet and educational attainment." That is an association across thousands of people, not a dose-response for you.
The intervention evidence is small. A 2021 pilot in Aging randomised "43 healthy adult males between the ages of 50-72" to an eight-week program of "diet, sleep, exercise and relaxation guidance, and supplemental probiotics and phytonutrients" or no intervention, and measured the 2013 clock in saliva. "The diet and lifestyle treatment was associated with a 3.23 years decrease in DNAmAge compared with controls (p=0.018)." Within the treated group the drop "by an average 1.96 years" came "with a strong trend towards significance (p=0.066)."
Read it with the authors' own caveat: "Larger-scale and longer duration clinical trials are needed to confirm these findings." Read it also knowing that the paper carries two published corrections, that the first-generation clock it used has the replicate noise described above, and that two authors declare they use the intervention in clinical practice and earn from related products. Forty-three men, eight weeks, a first-generation clock measured in saliva. Promising, and not a basis for buying anything.
What this means in practice: the behaviours that plausibly move a clock are the same behaviours that definitely move the markers below. You do not need the clock to justify them.
The markers with stronger mortality data
If the goal is a number that predicts how long and how well you live, and that moves when you train, several cheaper measurements have better evidence than any consumer clock.
Cardiorespiratory fitness
A 2018 cohort study in JAMA Network Open followed 122,007 patients through treadmill testing. The finding: "cardiorespiratory fitness was inversely associated with all-cause mortality without an observed upper limit of benefit." The size of the effect: "The increase in all-cause mortality associated with reduced cardiorespiratory fitness (low vs elite: adjusted HR, 5.04; 95% CI, 4.10-6.20; P < .001; below average vs above average: adjusted HR, 1.41; 95% CI, 1.34-1.49; P < .001) was comparable to or greater than traditional clinical risk factors."
Fitness is measurable with a treadmill test or, roughly, with a watch, and it moves in twelve weeks. The VO2 max protocol is the build.
Grip strength
The PURE study, published in The Lancet in 2015, measured grip in 139,691 people across 17 countries. "Grip strength was inversely associated with all-cause mortality (hazard ratio per 5 kg reduction in grip strength 1.16, 95% CI 1.13-1.20; p<0.0001), cardiovascular mortality (1.17, 1.11-1.24; p<0.0001)." And the comparison that should reorder your dashboard: "Grip strength was a stronger predictor of all-cause and cardiovascular mortality than systolic blood pressure."
The authors call it "a simple, inexpensive risk-stratifying method." A dynamometer costs less than one epigenetic kit and you can use it every month. The grip strength article has the norms and the training.
Resting heart rate, blood pressure, apoB
Three more rows, each with large cohort data behind it and each covered on this site: resting heart rate, which you can read every morning for free; blood pressure, which any pharmacy will measure; and apoB, the particle count that predicts heart attack risk better than LDL cholesterol and costs about as much as a standard panel.
None of these is "biological age." Together they are something better: a set of hard, repeatable numbers, each tied to mortality in studies of over a hundred thousand people, each of which responds to training, sleep and food within months.
The operator's annual dashboard
One page, one row per year, measured under the same conditions. Bring it to your doctor; do not act on it alone.
- Estimated VO2 max. From a supervised test if you can get one, otherwise the same watch and the same protocol each year. Direction matters more than the absolute figure.
- Grip strength. Dynamometer, both hands, best of three, same time of day. Monthly is fine; log the yearly best.
- Resting heart rate. Seven-morning average, measured before you sit up. The yearly row is the average of a normal week, not your best day.
- Blood pressure. Seated, rested, two readings, averaged. Pharmacy or home cuff, same device each year.
- ApoB and the standard panel. Annual blood work, same lab, with apoB added by request. Lipoprotein(a) once.
- Waist and bodyweight. Tape at the navel, morning, same day as the blood draw.
- Optional: an epigenetic clock. If you want one, treat it as a curiosity, not a KPI. Use a second-generation, principal-component version if offered, test under identical conditions, and judge only a multi-year trend. Never let it override the six rows above.
Then the Stoic test. The dichotomy of control says to spend your attention on what you can act on. Every row from one to six moves when you train before the business day, sleep a full night, eat like the Mediterranean studies and keep doing it for years. Row seven mostly measures noise you cannot act on. Build your discipline around the rows that respond to it.
The longevity for entrepreneurs article makes the full case for a short list of hard markers over a long list of expensive ones, and why that is a business decision as much as a health one.
Frequently asked questions
What is biological age?
Biological age is an estimate of how old the body is functionally rather than by the calendar. Epigenetic clocks estimate it from DNA methylation patterns: the 2013 multi-tissue clock used 353 CpG sites trained to predict chronological age across 8,000 samples, and later clocks such as PhenoAge and GrimAge were trained on clinical markers and mortality instead. The author of the first clock proposed it "measures the cumulative effect of an epigenetic maintenance system."
Are biological age tests accurate?
At population scale, second-generation clocks predict mortality well. For an individual, reliability is the problem. A 2022 Nature Aging study found technical noise produced deviations of up to 9 years between replicates of six prominent clocks, and a 2026 Aging Cell study found biological reliability across days was low to moderate for most clocks. A single consumer result should not be read as a verdict.
Can you reverse your biological age?
Not established at the individual level. One 2021 pilot trial in 43 men aged 50 to 72 reported a 3.23-year lower DNA methylation age after eight weeks of diet, sleep, exercise and relaxation guidance plus supplements, compared with controls, and its authors called for larger, longer trials. The behaviours involved reliably improve fitness, grip, resting heart rate and lipids, which have stronger mortality data than any clock.
What is the difference between first and second generation epigenetic clocks?
First-generation clocks were trained to predict chronological age, so their output is a guess at your birthday and the residual is the signal. Second-generation clocks such as PhenoAge and GrimAge were trained on clinical measures and time to death, and in their published validations outperform earlier clocks at predicting all-cause mortality, heart disease, cancer and other outcomes.
What markers predict longevity better than a biological age test?
Cardiorespiratory fitness: in 122,007 patients, low versus elite fitness carried an adjusted hazard ratio of 5.04 for all-cause mortality. Grip strength: in 139,691 people, each 5 kg reduction carried a hazard ratio of 1.16, and grip predicted mortality better than systolic blood pressure. Add resting heart rate, blood pressure and apoB, all repeatable, inexpensive and responsive to training.
Measure what moves, then move it
The epigenetic clocks are a genuine scientific achievement and their inventors describe them more cautiously than the companies that resell them. For one operator in one year, the number carries too much noise to steer by, and the researchers who built the clocks are the ones saying so.
What steers well is older and cheaper: how far you can run, how hard you can grip, how slowly your heart beats at rest, what your blood pressure and particle count read. Those numbers have been tested in hundreds of thousands of people, and every one of them answers to the same three disciplines that run the rest of the compound performance system. Train. Sleep. Eat like it matters. Repeat for a decade.
The free 5-Day Stoic Operator Challenge installs that rhythm in five days: training before the business opens, a daily review, and a structure built around what is in your control. Start the dashboard. Then start the work that changes it.
