Biological Age Clocks Compared: Phenotypic Age, GrimAge and DunedinPACE

What this page covers

Epigenetic clocks read chemical tags on DNA (methylation) from a blood sample and turn them into a number meant to reflect biological aging. Three names come up most: PhenoAge, GrimAge and DunedinPACE. This page compares how they were built and what studies found. There is no universal winner, and none is a diagnosis. Evidence labels: promising but limited, mixed, insufficient evidence, marketing claim. Education only; this page does not rank commercial tests.

The three clocks side by side

What each shows

PhenoAge. The authors reported it outperformed earlier measures for predicting a range of aging outcomes. [1] The paper notes that its authors are named inventors on a patent application for it. [1]

GrimAge. The authors reported it stood out among existing clocks for predicting time to death, heart disease and cancer in large validation data. [2] Because it includes a smoking estimate, part of what it captures is smoking history. [2]

DunedinPACE. The authors reported high test-retest reliability, associations with morbidity, disability and mortality, and effect sizes similar to GrimAge. [3] Some authors are listed as inventors on an invention licensed to a commercial entity. [3]

All three results are from observational cohorts of groups of people. An association with later illness in a group does not show that a score predicts one person's future or that changing the score changes the risk.

Do clocks change with interventions?

In the CALERIE randomized trial (220 adults without obesity, two years of calorie restriction versus usual eating), DunedinPACE showed a slowing of the pace of aging but PhenoAge and GrimAge did not change significantly, and treatment effects were small. [4] The authors said a conclusive test needs long-term follow-up on disease and death. [4] Different clocks can therefore tell different stories about the same people. A clock change is a biomarker change, not a health outcome.

Reliability and limits

• Technical noise. One study found technical noise could produce deviations of up to 9 years between replicate samples for six prominent clocks. A computational fix brought most replicates within 1.5 years. The study's authors disclosed ties to a company that licenses related metrics. [5]
• Statistical and validation challenges. Reviews say many computational and statistical challenges remain, [6] and that no consensus exists on how to validate aging biomarkers before clinical use. [7]
• Individual use. A 2025 critique argued that clocks fail to meet common standards for clinical utility compared with established biomarkers and that individual-level decisions based on them can be uninformative and potentially harmful. [8]

Label: promising but limited for group-level research; insufficient evidence for personal decisions.

Choosing between them

A researcher chooses based on the question: predicting mortality, estimating pace, or checking an intervention. For a reader, the practical point is that scores from different clocks are not interchangeable, so comparing a number from one test with another is not meaningful. See Direct-to-Consumer Epigenetic Age Tests, Biomarkers and Biological Age Tests and Epigenetic Clocks and DNA Methylation in Aging. Do not change medicines or supplements because of a clock result; see talking with a health professional.