Biological age is a real scientific idea wrapped in a consumer product category that outran the science. The idea is sound: two 50-year-olds can carry very different amounts of accumulated damage, and that difference predicts disease and death better than the birth certificate does. The product category is messier. Four different technologies all sell you a number, they are trained on different targets, and they routinely disagree about the same person by six years or more.
That leaves buyers with a specific problem. You paid $300, you got a number, and you have no way to tell whether the number is a signal, a lab artifact, or a reflection of how badly you slept the week you spat in the tube. This hub explains what each method measures, why they disagree, what actually moves the result, and how much movement is realistic.
The Verdict
The four methodologies, and what each one is actually reading
Every consumer biological age test uses one of four approaches. They are not interchangeable, and a number from one cannot be compared with a number from another.
| Method | Sample | Typical price | What it measures | Evidence status |
|---|---|---|---|---|
| DNA methylation clocks | Saliva or blood spot | $200–$500 | Chemical tags (methyl groups) at hundreds to thousands of CpG sites across the genome | Strongest research base. Second-generation clocks predict mortality better than chronological age alone. |
| Telomere length | Blood | $100–$300 | Average length of the protective caps on chromosome ends | Weak at the individual level. Measurement noise between draws often exceeds a decade of real change. |
| Biomarker composites | Standard blood panel | $0–$600 | Algorithms over routine labs: CRP, HbA1c, albumin, creatinine, white cell count, and similar | Transparent and actionable, but the number depends entirely on which formula the vendor chose. |
| Glycan age | Blood | $300–$500 | Sugar structures attached to circulating IgG antibodies | Tracks chronic inflammation well. Smallest published dataset of the four. |
Telomere testing deserves a specific warning. Telomere length varies substantially between cell types and between blood draws taken days apart. Published measurement error on consumer telomere assays can exceed the difference between a 40-year-old and a 55-year-old. It is the cheapest category to enter and the hardest to act on.
Not all methylation clocks are the same clock
"DNA methylation age" describes a family of algorithms, not one test. Two labs can run the same chemistry on the same saliva sample and report different ages because they run different clocks trained on different outcomes.
| Clock | Generation | Trained to predict | What that means in practice |
|---|---|---|---|
| Horvath (2013) | First generation | Chronological age | Trained to predict calendar age across tissues. Accurate at that job, which is why it responds weakly to lifestyle change. |
| PhenoAge (2018) | Second generation | Clinical health and mortality risk | Trained against nine clinical biomarkers plus age. Moves with metabolic and inflammatory status. |
| GrimAge (2019) | Second generation | Time to death, smoking history, plasma proteins | The strongest single predictor of mortality among common clocks. Heavily influenced by smoking exposure. |
| DunedinPACE (2022) | Pace-of-aging | Rate of aging per calendar year | Reports a speed (1.0 = one biological year per calendar year), not an age. Best suited to tracking an intervention. |
The practical takeaway is counterintuitive. A first-generation clock like Horvath is very accurate at guessing your calendar age — which is exactly why it is a poor progress tracker. If an algorithm was optimized to return your real age, it is partly designed to ignore the health differences you are paying to detect. Second-generation clocks and pace-of-aging measures were built to fix that, and they are the ones worth tracking.
What actually lowers biological age, ranked by evidence
Ranked by the strength of human evidence, not by how novel the intervention is. "Time to see change" is the interval before a matched retest is likely to show real movement.
| Intervention | Evidence | Why it moves the number | Time to see change |
|---|---|---|---|
| Stopping smoking | Strong | Largest single effect on GrimAge of any modifiable factor | 12–24 months |
| Correcting metabolic dysfunction | Strong | HbA1c, fasting insulin, and triglycerides feed PhenoAge directly | 3–12 months |
| Aerobic training (Zone 2 + VO2 max work) | Strong for mortality, moderate for clock movement | VO2 max is a stronger mortality predictor than most clocks | 6–12 months |
| Resistance training and protein intake | Strong for function, indirect for clocks | Preserves lean mass and insulin sensitivity | 6–12 months |
| Sleeping 7–9 hours consistently | Moderate | Short sleep raises CRP and glucose, both of which feed composite scores | 1–3 months |
| Reducing alcohol | Moderate | Improves liver markers, inflammation, and sleep architecture at once | 2–6 months |
| Treating chronic inflammation at its source | Moderate | CRP and IL-6 drive both glycan age and PhenoAge | 3–9 months |
| Caloric restriction or sustained fat loss | Moderate | The CALERIE trial showed a measurable slowing of DunedinPACE over two years | 12–24 months |
| Supplement stacks marketed for epigenetic age | Weak | Few randomized trials, small samples, short follow-up | Unproven |
| Plasma exchange and similar clinic protocols | Weak–emerging | Early human data, high cost, no hard outcome trials | Unproven |
Note the shape of the list. The top of it is unglamorous and mostly free; the bottom is expensive and unproven. That ordering is stable across every clock, because the second-generation clocks were trained on clinical biomarkers — so anything that improves glucose control, inflammation, liver function, and cardiovascular fitness improves the score almost by construction.
How much change is realistic
The most rigorous evidence available comes from CALERIE, a randomized trial of sustained caloric restriction. Two years of a serious, supervised intervention produced a small but statistically real slowing of DunedinPACE — a change in the rate of aging on the order of two to three percent, not a decade erased. Observational work on smoking cessation shows larger effects on GrimAge, because smoking status is baked heavily into that clock.
So set expectations accordingly. Someone who quits smoking, corrects prediabetes, drops significant visceral fat, and builds real aerobic capacity may see several years of improvement across two to four years of effort. Someone already lean, fit, non-smoking, and metabolically healthy has very little headroom — and should expect their number to barely move, because there is nothing left for the clock to detect.
There is also a failure mode specific to people who are already healthy. If your score comes back younger than your calendar age, the test gives you no direction. You cannot act on "keep doing what you are doing," and a second reading a year later will likely land within the noise band. For that reader the money is better spent on a VO2 max test, a DEXA scan, a coronary calcium score, or an ApoB measurement — all of which produce a specific target instead of a summary.
Should you test at all?
Testing is defensible in three situations. First, if you are running a defined multi-year intervention and want an aggregate outcome measure alongside your bloodwork. Second, if a single summary number is what motivates you to change behavior, and you know that about yourself. Third, if you are already deep in the data and the score is one input among many.
Skip it in two situations. If you have never run a comprehensive blood panel, that panel is strictly more useful and usually cheaper. And if you will only ever test once, the number has almost no meaning — biological age is a tracking tool, and a single reading with no comparison point tells you little that your fitness, waist circumference, and lipid panel do not already say.
Explainers
- What is biological age — and why chronological age misses the point
- How to lower biological age — what the evidence actually supports
- Methylation clock vs immune age — the two families of test, and what each is good for
- Genetic methylation test vs methylation clock: two different tests sold under one name
- How to read a biological age report — which number matters, and which is noise
- How to read a glycan biological age result — the five IgG indexes, and how slowly they move
- Average biological age by decade — why the average is always your chronological age, and what actually varies
- Does stress accelerate aging — what the evidence shows once behaviour is accounted for
Test reviews
- Best biological age test — five providers on one rubric
- Elysium Index review — DNA methylation, 9-marker epigenetic panel
- TruDiagnostic review — DNA methylation, DunedinPACE + TruAge algorithms
Related
- Biomarker guides — the underlying labs a biological age score summarizes
- Normal vs optimal ranges
- Reversal vs management — fixing the metabolic drivers that move these scores
- Supplement guides — what the evidence supports and what it does not
Frequently Asked Questions
Why do two biological age tests give me different numbers?
Because they measure different things and were trained on different outcomes. A methylation clock reads chemical tags on DNA; a biomarker composite runs a formula over your blood panel. Even two methylation tests disagree if one runs Horvath and the other runs PhenoAge, because Horvath was trained to predict calendar age and PhenoAge was trained to predict clinical health. A 6–10 year spread between vendors is common and does not mean either lab made an error.
Is biological age testing accurate?
Accurate at the population level, noisy at the individual level. Second-generation clocks like PhenoAge and GrimAge predict mortality across large groups better than chronological age does. But test–retest variation on the same person from the same sample is often 2–4 years, which is the same size as the change a year of serious effort produces. That gap is the central practical problem with the category.
How much can I actually lower my biological age?
Realistically 2–5 years across a multi-year, sustained intervention — and most of that comes from the largest, least glamorous levers: quitting smoking, fixing insulin resistance, losing visceral fat, and training aerobically. Marketing claims of 10 or 20 years reversed usually compare a first reading taken during illness, sleep deprivation, or an acute inflammatory episode with a later reading taken under better conditions.
Which biological age test should I choose?
If you want to track whether an intervention is working, choose a pace-of-aging measure such as DunedinPACE, which reports a rate rather than a number and is less sensitive to the baseline problem. If you want something actionable today, a standard blood panel interpreted against optimal ranges tells you more about what to change than any epigenetic score does, and usually costs less.
How often should I retest?
No more than once every 9–12 months, and always under matched conditions: same lab, same time of day, same fasting state, and not within two weeks of an illness, vaccination, hard training block, or a stretch of poor sleep. Testing quarterly mostly measures noise and the random state of your week.
Is biological age testing worth the money?
It is worth it if the number will change your behavior and you can afford to repeat it under matched conditions for several years. It is not worth it if you have never run a full blood panel — fasting insulin, ApoB, HbA1c, CRP, and a lipid panel cost less, are interpretable by any physician, and point at specific fixes. Buy the actionable data before the aggregate score.