Are Biological Age Tests Accurate? What the Clocks Can and Can't Tell You

Written by the Nuvirox Research Team

Short answer: accurate for what they were built for, unreliable for what most people buy them for. Epigenetic clocks are good at predicting outcomes across groups of people. They are considerably shakier at telling you whether your own number moved because of something you did, because the measurement noise between two draws can be as large as the effect you are hoping to see.

That gap — strong population prediction, weak individual tracking — is the single most important thing to understand before spending money on one.

Bar lengths are illustrative of predictive scope, not measured performance statistics. The key distinction is that later clocks were trained on health outcomes rather than on birth date.

What is a clock actually measuring?

DNA methylation — chemical tags attached to specific sites on the genome that change in patterned ways with age and exposure. A clock is an algorithm trained to read a subset of those sites and output a number.

Which sites, and what the algorithm was trained to predict, differs enormously between clocks. First-generation clocks were trained to guess chronological age from methylation. That is a strange target, because a perfect chronological age predictor would tell you nothing you did not already know.

Second-generation clocks were trained on health outcomes instead. This is why GrimAge outperforms the earlier clocks at predicting who dies sooner — it was built to do exactly that.

How is DunedinPACE different?

It measures a rate rather than a level. It was developed from the Dunedin cohort, in which 19 organ-system biomarkers were tracked repeatedly over roughly two decades, and trained on 173 methylation sites selected partly for their test–retest reliability.

A score of 1.0 means aging at one biological year per calendar year. Below 1.0 is slower, above is faster. This framing is more actionable than a static “your biological age is 47,” because a rate can plausibly change in response to something you do.

What human studies actually show

DunedinPACE was validated against hard outcomes, not just against age. Its development paper reported high test–retest reliability and associations with incident morbidity, disability, and mortality. Effect sizes were similar to GrimAge, and DunedinPACE added incremental prediction beyond it.

One randomized trial has moved a clock — barely. CALERIE randomized 220 adults without obesity to 25% caloric restriction or an ad libitum diet for two years. The restriction arm slowed on DunedinPACE by roughly 2–3%. PhenoAge and GrimAge did not change significantly. The authors interpreted this as evidence that pace-of-aging measures may be more sensitive to intervention than static biological-age estimates.

The honest counterweight is the reliability problem, and it comes from inside the field. Work on epigenetic clock reliability has distinguished technical reliability — how consistent the assay is if you run the same sample twice — from biological reliability, meaning how stable the measure is in the same person across ordinary exposures like diet, illness, and stress. Clocks that are technically excellent can still be biologically unstable, and researchers have developed principal-component-based versions specifically to address this problem for clinical trials and longitudinal tracking.

What a biological age test won't tell you

It will not diagnose anything. A concerning result is not a medical finding, and a reassuring one does not rule out disease. If you have symptoms, the appropriate response is a clinical workup, not a methylation panel.

It will not identify what to change. The number is a summary statistic. It cannot tell you whether the driver was sleep, alcohol, an infection three weeks ago, or the fact that your white blood cell composition shifted — which is itself a known influence on several clocks.

And it will not validate a supplement. No supplement has been shown in a randomized controlled trial to slow an epigenetic clock. Marketing that implies otherwise is running ahead of the evidence, including in the NAD+ category we work in.

If you are going to test anyway, how do you read the result honestly?

• Prefer a pace measure. DunedinPACE is designed for repeated measurement and was built around reliability, which is the specific weakness of the older clocks.
• Do not chase small changes. Given the retest variability documented in the reliability literature, a shift of a few percent between draws is not interpretable as a personal result.
• Space your tests widely. The only intervention shown to move a clock in a trial took two years to produce a 2–3% shift. Quarterly testing is measuring noise.
• Interpret alongside things that actually predict outcomes. Grip strength, walking speed, cardiorespiratory fitness, and standard blood work are cheaper, better validated, and directly modifiable — a point we make in our piece on VO2 max and everyday fatigue.

The functional measures deserve particular emphasis. Loss of muscle strength with age is a more consistent predictor of disability and death than loss of muscle mass, and it costs nothing to measure — a point we develop in our article on muscle loss and fatigue. That is a better use of attention than a methylation panel for most people.

It also helps to remember what a clock is a proxy for. Methylation change is one of twelve processes in the framework we cover in the hallmarks of aging, and no single one of those twelve summarises the others.

Frequently asked questions

Which biological age test is the most accurate?

For predicting mortality across populations, the second-generation and pace-of-aging clocks — GrimAge and DunedinPACE — outperform first-generation clocks. For tracking one person over months, none of them has established reliability that would justify strong conclusions.

Can I lower my biological age?

In the only randomized trial to test it, two years of 25% caloric restriction slowed one measure by 2–3% and did not move two others. That is the honest ceiling of what has been demonstrated, and it required a serious two-year intervention.

Why did my biological age go up after I improved my habits?

Most likely measurement variability. Retest noise, immune cell composition shifts, recent illness, and assay batch effects all influence results, and several of these can exceed the size of a genuine short-term change.

Are consumer epigenetic tests using the same clocks as research?

Some license validated algorithms; others use proprietary or reduced-site versions whose performance against health outcomes has not been independently established. It is worth asking a provider which specific clock they run and what it was validated against.

From Nuvirox

Why we formulated NAD+ Restore.

Every two-capsule serving delivers 500 mg of nicotinamide riboside chloride (NR) — one of the two most-researched NAD+ precursors, within the dose range used in published human trials.

It also includes 150 mg trans-resveratrol from Japanese knotweed and 50 mg quercetin from Sophora japonica — polyphenols studied alongside NAD+ pathways for cellular health support — plus 10 mg galactomannans from fenugreek to support absorption.

It ships with a 60-day money-back guarantee, which is long enough to actually evaluate it the way the research says you should.

Learn more about NAD+ Restore →

The bottom line

Epigenetic clocks are a real scientific advance and a poor personal dashboard. They earn their reputation on population-level prediction of who gets sick and who dies, and they lose it on the question most buyers actually care about: did my number move because of me? Until the reliability problem is solved, the fair reading is to treat a result as weak information, spaced years apart, interpreted alongside functional measures that are cheaper and better validated.

References

1. Belsky DW, Caspi A, Corcoran DL, et al. DunedinPACE, a DNA methylation biomarker of the pace of aging. eLife. 2022;11:e73420. PMCID: PMC8853656.
2. Waziry R, Ryan CP, Corcoran DL, et al. Effect of long-term caloric restriction on DNA methylation measures of biological aging in healthy adults from the CALERIE trial. Nat Aging. 2023;3(3):248–257. doi:10.1038/s43587-022-00357-y. PMCID: PMC10148951.
3. López-Otín C, Blasco MA, Partridge L, Serrano M, Kroemer G. Hallmarks of aging: an expanding universe. Cell. 2023;186(2):243–278. doi:10.1016/j.cell.2022.11.001. PMID: 36599349.
4. Higgins-Chen AT, Thrush KL, Wang Y, et al. A computational solution for bolstering reliability of epigenetic clocks: implications for clinical trials and longitudinal tracking. Nat Aging. 2022;2:644–661.
5. Mitchell WK, Williams J, Atherton P, Larvin M, Lund J, Narici M. Sarcopenia, dynapenia, and the impact of advancing age on human skeletal muscle size and strength; a quantitative review. Front Physiol. 2012;3:260. PMCID: PMC3429036.

These statements have not been evaluated by the Food and Drug Administration. This product is not intended to diagnose, treat, cure, or prevent any disease. This article is for informational purposes only and is not a substitute for professional medical advice.