G
Genetic Archaeology
Knowledge

Horvath Clock vs DunedinPACE Which Biological Age Test Wins

Published // November 22, 2026

You just dropped a few hundred bucks on a saliva kit, mailed it off, and got a PDF telling you that you are biologically 42 when your driver’s license says 38. Cool story. Now what? Most people stare at that biological age number and have zero clue if the Horvath clock or the newer DunedinPACE algorithm is the one they should actually trust. I have seen enough raw methylation reports to know that the headline number is usually the least useful part of the data.

What These Clocks Actually Measure

Forget the marketing fluff for a second. Every epigenetic clock relies on DNA methylation patterns at specific CpG sites. Think of methylation as rust accumulating on your genome. The Horvath clock, published back in 2013, was the first to show you could predict chronological age across almost any tissue using just 353 markers. It was a statistical miracle. But it was trained to match your calendar age, not your health trajectory. That is a massive distinction nobody explains on the sales page.

Horvath Clock The Original Pan Tissue Predictor

The original Horvath pan-tissue clock is brilliant at what it was designed for: estimating chronological age from a DNA sample. If you find a blood stain at a crime scene, this tells you roughly how old the suspect is. For longevity hacking? It is blunt. It correlates weakly with mortality risk compared to the newer models. Later versions like GrimAge and PhenoAge tried to fix this by training on plasma proteins and clinical biomarkers instead of just birth certificates. They are better, but they still output a static “age” number. A single snapshot tells you where you stood the day you spat in the tube. It does not tell you if you are aging fast or slow right now.

DunedinPACE The Speedometer You Actually Need

This is where DunedinPACE changes the game. Developed from the famous Dunedin Study cohort tracking 1,000 humans from birth to midlife, it does not guess your age. It estimates your pace of aging. The output is a ratio: 1.0 means you age one biological year per calendar year. 0.8 means you age slower. 1.2 means you are burning the candle at both ends. That is actionable. If your Horvath age is 50 but your DunedinPACE is 0.85, you are winning. If your Horvath age is 40 but your pace is 1.15, you have a fire to put out. Most direct-to-consumer reports bury this metric or leave it out entirely because a single “biological age” number sells kits better than a velocity vector.

Feature Horvath Clock (v1 / PhenoAge / GrimAge) DunedinPACE
Primary Output Static “Biological Age” (years) Rate of Aging (ratio/year)
Training Target Chronological Age / Clinical Biomarkers Longitudinal Physiological Decline
Sensitivity to Intervention Low (requires years to shift) High (detects change in months)
Best Use Case Forensics, Baseline Risk Stratification Tracking Lifestyle / Drug Efficacy
Tissue Specificity Pan-tissue (Blood, Saliva, Buccal) Optimized for Blood (Saliva is noisy)

Why Your Report Might Be Lying To You

Here is the dirty secret nobody wants to admit: saliva is a messy tissue. It contains epithelial cells and a variable mix of immune cells. Blood is the gold standard for DunedinPACE because the algorithm was trained on whole blood methylation. If you use a cheek swab or saliva tube, the cell-type composition shifts every time you brush your teeth or fight a cold. That noise can swing your “biological age” by five years between Tuesday and Friday. I have seen clients panic over a GrimAge jump that was just a mild viral infection. If you are serious about tracking, you need venous blood draws, not spit kits. And you need to test at the same lab, same time of day, same fasted state. Consistency beats the brand of the clock every single time.

💡 Practical Tip: Don’t test quarterly. Methylation changes slowly. Retest DunedinPACE every 6 to 12 months after a major protocol change (new exercise regimen, metformin, rapamycin trial). Testing sooner just measures noise and wastes money.

Cost is another factor the marketing ignores. A proper blood-based DunedinPACE panel from a CLIA lab runs $300 to $500. The spit kits are cheaper but significantly less reliable for pace tracking. If budget is tight, spend it on one high-quality blood draw per year rather than four noisy saliva kits. You also need to consider that your circadian rhythm impacts cortisol and methylation sampling; if you are a night owl forced into an 8 AM blood draw, your stress markers skew the data. We covered how CLOCK and PER2 genes dictate that rhythm in a previous deep dive.

Making Sense Of Your Biological Age Number

Stop chasing a lower number on a PDF. Chase a lower pace of aging. DunedinPACE is currently the only metric validated to respond to lifestyle interventions in human trials within a reasonable timeframe. Horvath and GrimAge are excellent for population risk stratification – they tell you if you are statistically likely to hit 80 or 90 – but they are terrible for N-of-1 biohacking feedback loops. Use the static clocks once as a baseline sanity check. Use the pace clock to steer the ship. And please, ignore the “biological age 23” result from that $150 saliva kit you bought on sale. You are not 23. The algorithm just overfit on a small CpG panel. Trust the velocity, not the snapshot.

Keep Reading

Related Research Guides

← Back to All Guides