Epigenetic Clocks: What Biological Age Tests Measure

This content is educational. It describes what biological age tests measure. It is not medical advice or a guide to using any specific test.

Why this matters

"Biological age" has become one of the most-marketed concepts in consumer longevity. Companies offer mail-order tests promising to reveal how fast you're "actually" aging. Influencers report their epigenetic age proudly. Wellness brands tie their products to the implicit promise of reversing it.

Underneath the marketing is a genuine scientific advance: the discovery that specific chemical modifications to DNA — methylation patterns — change in highly predictable ways with age. Algorithms that read these patterns can estimate chronological age with surprising precision. Some variants of these algorithms appear to predict not just chronological age but health outcomes and mortality.

The science is real and interesting. The consumer interpretation is more complicated. This page describes what epigenetic clocks measure, what they have been shown to predict, and where the limits sit.

What DNA methylation is

DNA methylation is a chemical modification — the addition of a methyl group (CH3) to a cytosine base in DNA — that affects whether and how genes are expressed. Methylation patterns are part of the epigenome, the layer of chemical modifications that determine gene activity without changing the underlying DNA sequence.

Methylation patterns are tissue-specific (different cell types have different patterns), responsive to environment (smoking, diet, and stress all influence methylation), and remarkably consistent across individuals at specific positions in the genome. Crucially, some methylation patterns shift in predictable ways with age.

The first generation: Horvath and Hannum clocks (2013)

In 2013, two papers transformed the field. Steve Horvath at UCLA and Greg Hannum at UCSD independently developed algorithms that used DNA methylation at specific sites in the genome to predict chronological age.

The Horvath clock uses methylation at 353 CpG sites across multiple tissue types and predicts chronological age with a median absolute error of approximately 3.6 years [Horvath 2013]. The Hannum clock uses 71 sites measured from blood and predicts age with similar precision [Hannum et al. 2013].

These first-generation clocks were optimised to predict chronological age — the calendar age of the participant. They were trained on large datasets of individuals with known birth dates.

A few features distinguished them as research breakthroughs:

• They worked across multiple tissues (Horvath)
• The accuracy was substantially better than previous methods
• The few people whose predicted age was very different from their chronological age were not random — they had systematic differences in health outcomes

This last finding launched the next generation of clocks.

The second generation: PhenoAge and GrimAge

Researchers realised that the more interesting question wasn't "how well can we predict chronological age" but "what predicts health outcomes." If methylation patterns are slightly different from someone's chronological age, do those differences mean something?

Several second-generation clocks emerged, designed not to predict birth date but to predict aging-related health outcomes:

PhenoAge (Levine 2018) — trained against a composite of clinical biomarkers (albumin, creatinine, glucose, CRP, lymphocyte percent, mean cell volume, red blood cell distribution, alkaline phosphatase, white blood cell count) that themselves predict mortality and morbidity. PhenoAge predicts both mortality and incidence of multiple age-related diseases more accurately than the first-generation clocks [Levine et al. 2018].

GrimAge (Lu 2019) — trained against time-to-death directly. Uses methylation patterns associated with specific plasma proteins linked to mortality and incorporates smoking pack-years. GrimAge currently predicts all-cause mortality more accurately than either PhenoAge or the first-generation clocks [Lu et al. 2019].

These second-generation clocks predict the same age in years but interpret it differently — they're answering "given your methylation, what is your expected health trajectory."

The third generation: DunedinPACE and rate-of-aging measures

In 2022, researchers published DunedinPACE — a clock that estimates the rate of biological aging rather than the accumulated age [Belsky et al. 2022].

DunedinPACE was trained on participants from the Dunedin Study (a 50-year longitudinal cohort in New Zealand) using 19 biomarkers measured repeatedly across decades. The output isn't a year value — it's a multiplier. A DunedinPACE of 1.0 means biological aging is occurring at the standard rate; 1.2 means aging is occurring 20% faster than average; 0.85 means 15% slower.

Conceptually this is an important shift. Earlier clocks measure "how much aging has accumulated"; rate-of-aging measures answer "how fast is aging happening right now." For interventions intended to slow aging, the rate metric is more directly informative.

What the clocks actually predict

Validation studies have established what each generation of clock has been shown to predict and at what strength:

First-generation clocks (Horvath, Hannum): - Chronological age with ~3-5 year precision - Modest association with mortality - Modest association with age-related disease incidence

Second-generation clocks (PhenoAge, GrimAge): - All-cause mortality (GrimAge strongest) - Cardiovascular disease incidence - Cancer incidence - Cognitive decline trajectories - Many other age-related outcomes [Higgins-Chen et al. 2022]

Third-generation clocks (DunedinPACE): - Current rate of physiological decline - Responsive to interventions in shorter time windows - Validated for measuring change rather than cross-sectional age

The pattern: each generation predicts something different. Marketing that conflates them — "my biological age is X according to my test" — often understates this complexity.

What consumer biological age tests measure

Several companies now offer mail-order epigenetic age tests. Common offerings:

• TruDiagnostic — uses extended versions of PhenoAge, GrimAge, and DunedinPACE
• Elysium Index — uses a proprietary methylation panel
• myDNAge — uses the Horvath clock
• Tally Health — uses a proprietary clock and tests other biomarkers

These tests typically require a saliva sample, a finger-prick blood sample, or a fingerstick dried blood spot. Cost ranges from approximately $200-500 per test.

What the test actually measures is methylation at a set of CpG sites. The clock algorithm is applied to those measurements to produce the biological age (or rate) estimate. Different companies use different clock variants, which is why the same person can get different "biological age" results from different tests.

Validation studies of these consumer tests have generally found that:

• The methylation measurements themselves are reproducible (within typical assay error)
• The clock outputs are reproducible when the same clock is applied to the same data
• Cross-test comparison is unreliable — different clocks produce different numbers

What the tests don't tell you

Several common misconceptions deserve explicit clarification:

Biological age is not deterministic. A high biological age is associated with elevated risk of age-related disease and mortality. It doesn't mean a specific outcome at a specific time.

Single test results have substantial measurement variability. A "biological age" that differs from chronological age by 1-2 years on a single test could reflect real biology or could reflect measurement noise.

Changes over time are more informative than single measurements. The literature on biological age decreasing in response to interventions (lifestyle change, supplementation) typically requires multiple measurements over months to be reliably detected.

Different clocks predict different things. A "high" PhenoAge means something different from a "high" GrimAge. The clocks aren't interchangeable.

The tests are not diagnostic. They are not validated for clinical decision-making and aren't typically used in standard medical care. The FDA has not cleared any biological age test for diagnostic claims.

What interventions affect biological age in research

The intervention literature on biological age clocks is growing rapidly. As of 2026, several patterns are emerging from published trials:

• Caloric restriction — the CALERIE trial measured DunedinPACE in participants on 25% caloric restriction for 2 years; intervention participants showed approximately 2-3% slower aging rate than controls [Waziry et al. 2023]
• Exercise — multiple cohort and intervention studies have reported slower epigenetic aging in physically active populations [Sillanpää et al. 2018]
• Diet quality — Mediterranean diet patterns and similar are associated with slower epigenetic aging across cohorts
• Specific supplements — some small trials have reported epigenetic age effects from various supplements, though replication has been mixed
• Rapamycin — animal studies show clear effects; human trials are ongoing

The translation question — whether moving a clock value actually translates to better long-term health — is still being established. Several trials are tracking participants long-term to confirm.

What Proco's editorial position is

Epigenetic clocks represent genuine scientific progress in measuring biological aging. The technology behind them is well-validated, and the second- and third-generation clocks predict health outcomes that earlier methods couldn't.

The consumer market around them has expanded faster than the validation. Confident claims about "reversing your biological age" or "biological age X equals lifespan Y" usually overstate what the underlying research supports.

For readers considering biological age testing: the information may be interesting, particularly for tracking trends over multiple tests. It is not necessary for making sensible health decisions, and the specific number should not be taken as more precise than the science currently supports.

For supplement-based interventions claiming to "slow biological aging": the published evidence varies dramatically by intervention. The compliance line we use is to describe what the research has measured for the specific intervention, not to translate it as "this reverses aging."

Related Proco pages

• The hallmarks of aging
• Healthspan vs lifespan
• Why evidence-based gets misused

Sources

Proco provides educational, research-based information. This page describes what biological age tests measure. It is not medical advice or a recommendation for any specific test. For health decisions, consult a qualified healthcare professional.

Schema (for implementation)

{
  "@context": "https://schema.org",
  "@type": "Article",
  "headline": "Epigenetic Clocks: What Biological Age Tests Measure",
  "description": "Epigenetic clocks use DNA methylation patterns to estimate chronological age, biological age, and rate of aging. This page describes the three generations of clocks and what each actually predicts.",
  "datePublished": "2026-06-01",
  "dateModified": "2026-05-31",
  "author": {"@type": "Organization", "name": "Proco"},
  "publisher": {"@type": "Organization", "name": "Proco", "url": "https://procohq.com"},
  "about": {"@type": "Thing", "name": "Epigenetic clocks and biological age"}
}