What a biological age test actually measures

You can now spit into a tube, mail it off, and get back a number that claims to be your real age. Not the one on your license, the one your cells are supposedly living at. The kits sell the gap between the two as the whole point: you're 44 on paper but 39 underneath, or the reverse, and either way there's a chart with your dot on it. It's a tidy story. The thing being measured is more interesting than the story, and a little less obedient.

What the tube is actually reading

These tests don't read your DNA the way an ancestry kit does. They read what sits on top of it. Scattered across the genome are sites where a small chemical tag, a methyl group, can attach to the DNA and quietly change how a gene gets used. The pattern of those tags is called the methylome, and it drifts in surprisingly regular ways as a body ages. Some sites pick up tags over the decades, others lose them, and the drift is consistent enough across people that you can build a statistical model around it.

That model is the "clock." Back in 2013, Steve Horvath showed you could take methylation readings from a few hundred of these sites, weight them, and predict someone's calendar age within a few years. The test isn't sensing damage or vitality directly. It's noticing that your tags now look like the tags of people who are, on average, a certain age, and reporting that age back to you.

Why two clocks disagree about the same person

Here's the part the marketing flattens. There isn't one clock. There are many, and they were built to answer different questions.

The first generation, Horvath's among them, were trained to guess chronological age. They're good at it, which is exactly their limit: a tool optimized to match the number on your license isn't really telling you anything the license didn't. The more interesting clocks came next. PhenoAge and GrimAge were trained not against the calendar but against health markers and mortality data, so they lean toward how a body is faring rather than how long it's existed. A newer approach, DunedinPACE, skips the age question entirely and estimates a rate: how fast you appear to be aging right now, in biological years per calendar year.

Because they're built on different targets, they can disagree about you. One clock can call you younger than your age while another calls you older, on the same vial of saliva. That isn't a malfunction. It's three instruments measuring three related but distinct things and being honest about it.

The noise underneath the number

Even a single clock isn't as steady as the clean readout suggests. Methylation measurement carries real technical noise, enough that running the same sample twice can shift the result by a year or two with nothing about you having changed. A reading also captures one morning, not a trajectory. The pace-of-aging clocks were partly designed to address that, but a lone measurement is still a snapshot, and a snapshot can't tell you which direction you're moving.

This matters because the tests are research instruments, validated across large populations, repackaged as personal verdicts. A clock can be genuinely useful for studying whether an intervention nudges aging across thousands of people, which is how the recent semaglutide work used them, and still be a blunt tool for telling one person what to do on a Tuesday.

How to hold the result

A biological age number is a real measurement of a real thing. It is not a diagnosis, a forecast, or a score to beat. The most defensible way to read one is as a low-resolution reflection, interesting in aggregate and over time, close to meaningless as a single dramatic figure. If you ever take one, the second reading months later tells you more than the first one ever could, and even then it's a hint, not a ruling.

The honest version of the pitch is quieter than the one on the box. Aging leaves marks you can now partly read. Reading them is not the same as controlling them, and a number is not the same as knowing.

---

This post is educational and general in nature. It is not medical advice. For guidance about your own health, talk to a qualified clinician.