What Epigenetic Clocks Measure

Every cell in your body carries the same DNA sequence, yet a liver cell behaves nothing like a neuron. The difference lies in the epigenome — a layer of chemical tags that sit atop the DNA and control which genes are switched on or off without changing the underlying sequence. The most studied of these tags is DNA methylation: the addition of a methyl group to a cytosine nucleotide, almost always when it appears before a guanine in what's called a CpG dinucleotide site.

As we age, these methylation patterns change in highly predictable ways. Some sites gain methylation; others lose it. The pattern is consistent enough across individuals that a mathematical model can read the methylation status of a subset of CpG sites and estimate a person's age with remarkable accuracy. These models are epigenetic clocks.

The CpG Methylation Mechanism

The human genome contains approximately 28 million CpG sites, but only a fraction change meaningfully with age. Epigenetic clocks identify the most informative subset — sites where methylation tracks age in a linear, reproducible fashion — and build a weighted algorithm to produce an age estimate in years. The output is called epigenetic age, DNA methylation age, or biological age, depending on the clock and the context.

What makes this powerful is that the same underlying biology — the enzymatic machinery that adds and removes methyl groups — is disrupted by exactly the lifestyle factors we associate with poor health: smoking, obesity, chronic stress, and sedentary behavior. Conversely, the same machinery is supported by the factors associated with healthy aging. The epigenome is not just a passive record; it is a live readout of biological wear and tear.

First-Generation Clocks: Horvath and Hannum

Steve Horvath, a biostatistician at UCLA, published the first pan-tissue epigenetic clock in 2013 using 353 CpG sites selected from 8,000 blood and tissue samples. The Horvath clock predicted age within a median absolute deviation of 3.6 years across 51 tissue types — a feat of biological universality that immediately attracted attention. Its pan-tissue nature suggests it captures something fundamental about the aging program rather than tissue-specific drift.

In the same year, Gregory Hannum and colleagues published a blood-specific clock using 71 CpG sites that performed similarly in blood but didn't generalize across tissues. The Hannum clock correlates more strongly with physiological measures like telomere length.

Second-Generation Clocks: PhenoAge and GrimAge

First-generation clocks are good at predicting chronological age. But chronological age is not what we ultimately care about — we care about mortality risk, healthspan, and organ function. Second-generation clocks were trained not on age alone but on clinical outcomes.

Morgan Levine's PhenoAge (2018) used a composite clinical biomarker score (creatinine, albumin, glucose, CRP, lymphocyte percentage, MCV, red cell distribution width, alkaline phosphatase, and white blood cell count) as its training target, producing a clock that better predicts mortality, morbidity, and age-related diseases than any first-generation clock.

Lu et al.'s GrimAge (2019) went further, training on plasma protein proxies for age-related phenotypes. GrimAge consistently outperforms all earlier clocks at predicting time to death, time to coronary heart disease, and lung function — earning its grim name for good reason.

Third-Generation: DunedinPACE

The Dunedin Study in New Zealand tracked a birth cohort from birth to age 45, collecting biomarkers at every wave. DunedinPACE (Belsky et al., 2022) uses 173 CpG sites not to estimate an age in years but to estimate the speed of aging — how many biological years a person ages per calendar year. A DunedinPACE score of 1.0 means you are aging at the average rate. A score of 1.2 means you are aging 20% faster than average. This shift from biological age estimate to pace of aging has proven more sensitive to short-term interventions and lifestyle changes.

How Methylation Changes with Age

The aging methylome does not change randomly. Decades of research have revealed two overarching patterns that run simultaneously and largely independently of each other.

Gain of Methylation at Polycomb Regions

Polycomb group proteins maintain embryonic stem cell pluripotency by silencing developmental genes. As cells age, the target sites of these proteins — called polycomb repressive complex (PRC2) targets — progressively gain methylation. This means that many of the genes important for cell fate specification and tissue maintenance become increasingly silenced with age. The effect is likely a contributor to the reduced regenerative capacity observed in aged tissues.

The Horvath clock sites are enriched for these polycomb targets, suggesting that the clock is partly reading this epigenetic silencing process. The question of whether this silencing is causally involved in aging or is merely a biomarker of it remains under active investigation.

Loss of Methylation at CpG Islands

Running in the opposite direction, CpG islands — dense clusters of CpG sites in gene promoters, where methylation typically represses transcription — tend to lose methylation with age in some tissues. This phenomenon, sometimes called epigenetic drift, may contribute to the inappropriate activation of genes that should remain silent, including retrotransposons and genes associated with inflammation.

Drift vs. Programmed Change: The Open Debate

One of the central controversies in the field concerns whether epigenetic aging is a programmed process — an active developmental program that continues past reproductive maturity — or stochastic accumulation of damage and maintenance failure. Evidence for programming comes from the observation that clocks work pan-tissue (suggesting a systemic signal) and from parabiosis experiments showing that old mice can be rejuvenated by young blood factors. Evidence for drift comes from the observation that methylation becomes increasingly heterogeneous between cells as organisms age, consistent with random errors in maintenance methylation after every cell division.

Key insight: The programmed-vs-drift debate matters practically. If aging is programmed, it may be reversible by resetting the program (as Yamanaka factor experiments suggest). If it is predominantly drift, interventions must focus on reducing the rate of methylation errors — through antioxidants, NAD+ support, and reducing replication stress.

What Accelerates Your Epigenetic Clock

The single most important practical insight from epigenetic clock research is that lifestyle and environment move the needle — sometimes dramatically. Below are the best-quantified accelerators, with approximate effect sizes drawn from large meta-analyses and population studies.

Smoking

Tobacco smoking is the most studied epigenetic accelerant. Multiple large studies have quantified the effect at 4–9 biological years of acceleration, with the magnitude depending on pack-years and the specific clock used. Crucially, methylation at smoking-associated sites begins to reverse after cessation — but reversal is incomplete and can take decades for heavily exposed individuals. Some smoking-associated CpG sites retain their altered methylation status even 30+ years after quitting, encoding a kind of biological memory of the exposure.

Obesity

Body mass index above 30 is associated with approximately 2.2 years of GrimAge acceleration in large cohort studies. The mechanism is likely multifactorial — chronic low-grade inflammation driven by adipose tissue, metabolic dysregulation affecting one-carbon metabolism (which directly feeds into methylation), and sleep disruption secondary to sleep apnea. Weight loss through caloric restriction or bariatric surgery has been shown to partially reverse this acceleration.

Chronic Psychological Stress

Post-traumatic stress disorder, caregiver burden, and high perceived stress scores are each associated with approximately 1–2 years of epigenetic age acceleration. A 2019 study of Holocaust survivors found detectable epigenetic aging signals compared to unexposed controls of the same chronological age. The glucocorticoid hypothesis proposes that chronically elevated cortisol disrupts DNMT3A and TET enzyme activity, altering the methylation landscape at stress-responsive loci.

Poor Sleep

Habitual short sleep duration — less than 6 hours per night — is associated with roughly 1.5 years of epigenetic age acceleration on PhenoAge and DunedinPACE. Shift workers show even larger effects. Sleep appears to be a period of epigenomic maintenance; chronic sleep restriction likely impairs this process.

Sedentary Behavior

The contrast between highly active individuals (≥7,500 steps/day, regular vigorous exercise) and sedentary controls produces a difference of up to 1.8 years on GrimAge. Even in older adults, initiating a regular walking program produces measurable reductions in epigenetic age within 12–16 weeks. The effect appears to be driven by improved mitochondrial function, reduced inflammatory signaling, and enhanced insulin sensitivity — all of which interact with the methylation machinery.

Diet Quality and Ultra-Processed Foods

High intake of ultra-processed foods is associated with accelerated PhenoAge independent of total caloric intake. Conversely, Mediterranean diet adherence is associated with epigenetic age deceleration, with some studies reporting 1–3 year differences between high and low adherence groups. Specific dietary factors with epigenomic evidence include folate (essential for one-carbon metabolism and methylation capacity), B12, and polyphenols, which modulate DNMT and TET enzyme activity.

Interventions That Reverse Epigenetic Age

The most exciting development in epigenetics over the past decade is the demonstration that biological age, as measured by epigenetic clocks, can be reversed — not just slowed, but reversed. Here is the current evidence hierarchy.

The TRIIM Trial: Proof of Principle for Reversal

The TRIIM (Thymus Regeneration, Immunorestoration, and Insulin Mitigation) trial, published by Greg Fahy and colleagues in 2019, enrolled nine healthy men aged 51–65 and administered a combination of recombinant human growth hormone (0.015 mg/kg/day), metformin (500 mg twice daily), and DHEA (50 mg/day) for 12 months. Epigenetic age was measured using four clocks at multiple time points.

All participants showed a reduction in epigenetic age. The average reversal across clocks was 1.9 years — and crucially, the reversal continued after the intervention ended. At the 6-month follow-up, participants had maintained their younger epigenetic age. While the trial was small and uncontrolled, it was the first human demonstration of epigenetic age reversal and ignited intense interest in the field. A larger, controlled TRIIM-X trial is ongoing.

Partial Cellular Reprogramming: Yamanaka Factors

The most striking preclinical results come from partial reprogramming using Yamanaka transcription factors (Oct4, Sox2, Klf4, c-Myc). Full reprogramming reverts somatic cells to induced pluripotent stem cells and erases their epigenetic age entirely — but also erases cell identity, which would be incompatible with use in a living organism. Partial or transient reprogramming, using pulses of these factors, has been shown in multiple mouse studies to rejuvenate epigenetic age while preserving cell identity. Researchers at Altos Labs, the Salk Institute, and elsewhere are working to translate this approach to human therapeutics. The timeline to clinical application remains uncertain, but this represents the most radical biological age reversal mechanism currently under study.

Caloric Restriction

The CALERIE trial — a randomized controlled trial of 25% caloric restriction in humans — found that two years of restriction produced a significant reduction in DunedinPACE (by approximately 0.02 units, or roughly a 2–3% reduction in aging pace) compared to controls, as reported in the post-hoc epigenetic analysis published in 2023. This is the largest RCT evidence base for dietary epigenetic age modification in humans. The effect was mediated in part by reductions in systemic inflammation and improvements in metabolic markers.

Aerobic Exercise

High-intensity interval training (HIIT) in older adults has been shown to reduce GrimAge by up to 1.3 years over 26 weeks in randomized trials. Endurance exercise exerts epigenomic effects through multiple pathways: lactate-dependent epigenetic signaling, PGC-1α-mediated mitochondrial biogenesis, and reductions in pro-inflammatory cytokines that disrupt methylation homeostasis. The effect is dose-dependent, with vigorous exercise showing larger effects than moderate-intensity work.

Metformin

Metformin, the most widely prescribed diabetes drug, activates AMPK and has shown life-extension effects in multiple model organisms. In humans, diabetic patients on metformin age more slowly than non-diabetic controls (by epigenetic measures) — a paradoxical finding given that diabetes itself accelerates aging. The TAME (Targeting Aging with Metformin) trial is the first FDA-approved clinical trial using aging itself as a primary endpoint, with epigenetic clocks as secondary outcomes. Preliminary data from the TRIIM trial suggests metformin's contribution to the epigenetic reversal observed may be partly through insulin sensitization that allows GH to exert its effects without inducing IGF-1-related insulin resistance.

Fasting Protocols

Prolonged fasting (3–5 days) activates autophagy pathways and has been associated with reductions in markers of cellular aging, though direct epigenetic clock data from human fasting trials are still limited. A 2021 study of a 5-day fasting-mimicking diet protocol showed reductions in biological age (on the Klemera-Doubal biological age algorithm) that persisted for months after the fast. Intermittent fasting studies with epigenetic clock outcomes are ongoing.

Rapamycin

Rapamycin (sirolimus) inhibits mTORC1 and is the most consistently life-extending compound across model organisms. In mice, rapamycin started late in life extends median lifespan by 10–15%. In dogs, a recent small trial showed improved cardiac function. Human data on epigenetic clocks is emerging from compassionate use and research contexts. The dose, timing, and intermittent vs. continuous protocols that maximize benefit while minimizing immunosuppressive side effects remain under active investigation. Many longevity researchers now take low-dose intermittent rapamycin off-label based on the preclinical evidence.

Mediterranean Diet

Multiple cohort studies and the PREDIMED randomized trial have demonstrated that high Mediterranean diet adherence is associated with reduced epigenetic aging on multiple clocks. The dietary pattern provides high folate (methylation substrate), polyphenols (TET enzyme activators), omega-3 fatty acids (anti-inflammatory epigenomic effects), and low ultra-processed food intake. The effect sizes in observational studies range from 1 to 3 years of epigenetic age advantage in high versus low adherence groups.

Ready to build the full protocol?
The Longevity Stack ranks 20+ compounds by evidence tier — the full biomarker panel, VO2 max protocol, sleep science, hormone chapter, and three budget-level stacks, built from the same research on this page.
Get the Longevity Stack → $19

Testing Options and Interpretation

Commercial epigenetic age testing has matured rapidly since 2020. Three companies currently dominate the consumer market, each with distinct methodologies, reporting formats, and clock selections.

TruDiagnostic (TruAge Complete)

TruDiagnostic's flagship TruAge Complete test measures multiple clocks simultaneously — Horvath, Hannum, PhenoAge, GrimAge, and DunedinPACE — from a single blood spot sample collected at home. The company uses Illumina's EPIC array (850K CpG sites), which is the gold standard platform for methylation research. Reports include your biological age on each clock, an immune age estimate, a telomere length estimate, and a DunedinPACE score. Cost is approximately $299 per test. TruDiagnostic's research partnerships and transparent methodology have made it the preferred option for many longevity researchers using self-tracking.

Elysium Index

Elysium Health's Index test measures a proprietary single-clock biological age from blood spot samples. The company collaborated with Morgan Levine (creator of PhenoAge) in developing their algorithm. Index is available by subscription ($249/year for one test, or less per test with a higher-tier subscription) or as a single test at $299. The reporting is clean and consumer-friendly but provides less granularity than TruDiagnostic's multi-clock output.

MyDNAge

MyDNAge uses either blood or urine samples (the urine option is a unique convenience differentiator) and reports primarily on the Horvath and Hannum clocks. At approximately $299 per test, the urine-based option is less invasive and attractive for frequent monitoring. However, urine-based methylation testing captures epithelial cell DNA, which may differ from blood-based readings, and longitudinal comparability between sample types should be interpreted cautiously.

What a Result Actually Means

Receiving an epigenetic age result that is younger than your chronological age is associated with lower mortality risk and better physiological function in population data. However, an individual result carries significant uncertainty. Test-retest variability for most commercial platforms is ±1.5 to 3 years, meaning that a 2-year improvement may be within measurement noise for a single individual. The signal becomes meaningful when you track multiple time points, when changes are large (greater than 4–5 years), or when multiple clocks move in the same direction.

Practical guideline: Don't make major medical decisions based on a single epigenetic age test. The value lies in tracking trends over time — ideally at 6-month intervals with the same platform — and using the result as one input in a broader health optimization picture alongside standard biomarkers, VO2max, grip strength, and clinical lipid panels.

Limitations of Current Testing