Biological Age vs. Chronological Age

You know your chronological age. It’s on your driver’s license, it goes up by one every birthday, and nobody has ever tried to sell you a kit to measure it. What a test kit, a clinic or a headline may have just told you is that you also have a biological age, and that it can be a different number. Humanic Blue is here to untangle what those two numbers do—and don’t—tell you.

Chronological age counts time since birth. Biological age is a model’s estimate from measurements such as blood markers or DNA methylation, the chemical tags on DNA that some clocks read. Different models answer different questions, and a younger-looking result does not establish that you will live longer. Biological age vs chronological age is a comparison between a calendar fact and an estimate—not two equally certain measurements.

Researchers have linked older-than-expected scores on several clocks to illness and earlier death in large groups. Those associations do not give one person a precise forecast or show that lowering a score helps people live longer and better. The method, its measurement error and what it was trained to predict all matter.

What does “biological age” actually mean, in one blunt sentence?

Biological age is a number calculated from measurements of your body using a model developed from other people’s data.

There’s no organ in your body that stores it. Every version of it is a model, and different models measure different things: DNA methylation (chemical tags on your DNA that shift predictably with age), proteins in the liquid part of blood, routine blood-test values, or a mix.

Because each model is trained on a different dataset with a different goal, they don’t agree with each other. That’s the first thing worth knowing before you take any one number personally.

Steve Horvath developed the 2013 Horvath clock at UCLA and subsequently worked at Altos Labs. Speaking to The Scientist, he put the state of the field plainly: “There’s no question that epigenetic clocks perform very well in epidemiological studies. But what the field has not yet achieved is really what one would call reference standards.” Translation: the clocks work on populations; nobody has agreed what a “correct” reading for one person would look like.

If chronological age is so crude, why do doctors still use it?

Because it remains the single strongest predictor of most chronic diseases.

The National Institute on Aging’s geroscience program describes aging as a major risk factor for multiple chronic diseases, frailty and reduced resilience. That does not mean a birthday predicts an individual’s future precisely.

Researchers use chronological age as a point of comparison when testing whether a biological-age model adds useful information. Morgan Levine, PhD, who developed PhenoAge and later ran labs at Yale and at Altos Labs, designed her clock so that its acceleration score has, in her words, “a correlation of r=0.0 with chronological age.” That acceleration score describes whether the model reads you as older or younger than expected for your calendar age. It is not a diagnosis or a direct measurement of how many years remain.

What does “age acceleration” mean, statistically?

Age acceleration sounds like something a sports car does. In these models, it means the difference between a clock result and the result expected for someone your calendar age.

Researchers draw a line through thousands of people plotting predicted age against real age, then measure how far above or below that line you fall. That distance, the residual, is your age acceleration. A positive score means the clock reads you as older than expected for your calendar age; a negative score means younger than expected.

Horvath’s original clock learned this line from 8,000 samples across 51 tissues. GrimAge later linked older-than-expected scores with earlier death in nearly 7,000 people. Both findings describe patterns across groups; neither converts one person’s result into years left to live.

The gap also contains measurement error. In 2022, Albert Higgins-Chen, Morgan Levine and colleagues found that laboratory noise could move six familiar clocks by as much as nine years between runs of the same sample. Their re-engineered versions brought more than 90% of repeats within about 1–1.5 years. If your result says +3, asking which version produced it is not nitpicking. It is the difference between a signal and a lab having a Tuesday. The clock statistics

Why do first-generation and second-generation clocks give different answers?

Because they were trained to predict different things. Same word, “clock,” very different jobs.

• First-generation clocks were trained to estimate chronological age from biological measurements.
• Second-generation clocks, including PhenoAge and GrimAge, incorporated disease-risk or mortality information.
• Pace measures, such as DunedinPACE, estimate a rate of change rather than an age.

The studies and training populations are detailed below.

Levine’s concern was simple: train a clock to guess birthdays and it may ignore biological signals that matter for disease but do not neatly follow the calendar. PhenoAge instead began with nine routine blood values plus age, chosen to distinguish illness and mortality risk among people of the same age. In validation cohorts, each extra DNAm PhenoAge year was associated with a 4.5% rise in mortality risk during follow-up.

So a first-generation clock asks, “How old do you look, molecularly?” A second-generation clock asks, “How much risk do you carry for your age?” Same dashboard word, different warning light. The training populations and results

How does a clock's training goal change the meaning of the age it reports?

How can two 45-year-olds be aging at 0.4 and 2.4 years per year?

The Dunedin Study followed about 1,000 New Zealanders born in 1972–73. Researchers measured 19 indicators across the heart, metabolism, kidneys, liver, immune system, teeth and lungs at ages 26, 32, 38 and 45. At 45, the slowest measured pace was 0.40 biological years per calendar year and the fastest was 2.44. Faster agers also tended to walk and balance less well and showed more cognitive decline. Same birthday; remarkably different mileage.

Daniel Belsky and colleagues then compressed those two decades of change into DunedinPACE, a blood test designed to measure speed rather than position. In roughly 2,500 Framingham participants, faster scores were associated with higher mortality, and repeat runs of the same sample agreed closely. Neither finding proves that lowering the score will extend life.

Belsky and several co-authors are inventors on technology licensed commercially, and NBC reports that Belsky occasionally advises TruDiagnostic, the exclusive licensee. That relationship does not cancel the cohort. It belongs beside the result. The cohorts, statistics and full disclosure

Where has the gap been validated at scale?

Researchers have tested these approaches in the Dunedin Study in New Zealand, US NHANES surveys, the Framingham Heart Study and UK Biobank. These are large research cohorts, not interchangeable samples of every person who might buy a test. The cohort-by-cohort evidence is in the resources below.

In all four, a gap between modelled age and calendar age predicts outcomes in groups. None was designed to hand one person a verdict.

What remains unanswered when an age gap predicts outcomes in groups rather than individuals?

Is “biological age” even one measurable quantity?

This is the counter-case, and it comes from inside the field.

In a 2024 survey, about 100 aging researchers—including Belsky, Horvath and Levine—gave substantially different answers about aging’s definition, causes and possible reversal. The Biomarkers of Aging Consortium has likewise identified the lack of agreed standards as a barrier to clinical validation.

Researchers Idan Shalev and Abner Apsley warn that reducing this complicated process to one number can mislead, especially when a research tool built to sort populations is used to describe an individual. João Pedro de Magalhães made the practical point to MIT Technology Review: five years younger on a clock does not necessarily mean five extra years of life. Younger on the dashboard is not time in the bank.