Omega-3 nudged three of four epigenetic aging clocks, by a few months over three years. That is the headline, and it is both more interesting and much smaller than it sounds. The result comes from a Swiss-led analysis of the DO-HEALTH trial, published online in February 2025, and it is a far larger randomized test than the small pilot studies that came before it. It is also a blood-test result in healthy 75-year-olds. It says nothing direct about how long anyone lives.
Who took part, and what they took. DO-HEALTH enrolled 2,157 generally healthy adults aged 70 and older across five European countries. They were randomized into eight arms of 264 to 275 people each, covering every combination of three treatments: 2,000 IU of vitamin D3 a day, 1 g of omega-3 a day (330 mg EPA plus 660 mg DHA, from marine algae), and a simple home strength program of 30 minutes, three times a week. Each treatment had a matching control: identical placebo capsules, or a joint-flexibility routine for the exercise arm. The aging analysis covers the 777 Swiss participants who consented and gave blood at the start and again after 3 years. Their average age was about 75, and roughly 60% were women.
What the omega-3 group showed. The researchers tracked four epigenetic measures: PhenoAge, GrimAge, GrimAge2 and DunedinPACE. Omega-3 alone moved three of them in the younger direction. For PhenoAge the difference was d = -0.16 in standardized units (95% CI -0.30 to -0.02), for GrimAge2 d = -0.32 (95% CI -0.59 to -0.06), and for DunedinPACE d = -0.17 (95% CI -0.31 to -0.04). The original GrimAge did not move, and neither did the two older first-generation clocks, Horvath and Hannum. In the authors' summary, the standardized effects ranged from 0.16 to 0.32 units, which they translate into 2.9-3.8 months over the 3 years for PhenoAge and GrimAge2.
How a methylation clock works. DNA carries small chemical tags called methyl groups, and the pattern of those tags shifts with age. An epigenetic clock is a formula that reads the tags at many spots across the genome and folds them into one number. The lab in this trial measured more than 850,000 sites per blood sample and ran the results through four formulas. They differ in what they were built to predict. The first-generation clocks were trained to guess calendar age, which they do well, but they say little about health. PhenoAge and GrimAge2 were trained on mortality risk, so they track who is more likely to fall ill or die. DunedinPACE reports speed instead of age: it was built from how fast 19 measures of organ-system health declined over two decades in a New Zealand birth cohort, and a value of 1 means one year of biological change per calendar year.
What three months actually means. It does not mean the average participant gained three months of life. It means the omega-3 group's estimated biological age rose about three months less than the placebo group's, on two clocks, over the same three years. DunedinPACE, the speed clock, came out about 1% slower by the authors' estimate. For scale, the CALERIE trial put 220 adults without obesity on a 25% calorie cut for two years and slowed DunedinPACE by what the DO-HEALTH authors describe as 2-3%, an effect CALERIE's own authors called small. Nobody can read a personal result into a group average like this, and the paper itself notes that there is no gold-standard measure of biological aging.
Vitamin D and exercise on their own. Neither moved a single clock when given alone. The only place the treatments seemed to add up was PhenoAge, where omega-3 combined with vitamin D, exercise or both produced differences between -0.24 and -0.32, against -0.16 for omega-3 by itself. GrimAge, GrimAge2 and DunedinPACE showed no such pattern. So omega-3 carried the signal, and the "all three together" story rests on one clock. It is also worth remembering that 88% of these participants were already physically active, and the exercise program was deliberately modest, so there may have been little room for it to show a difference.
Reasons to hold the result loosely. First, this is a post hoc analysis: the clock question was added to an existing trial, not planned as its primary goal, and the paper says so. Second, there was no single primary outcome and no formal correction for looking at four clocks; the authors looked for a consistent pattern with confidence intervals that excluded zero, and the PhenoAge interval only just does (its edge is -0.02). Third, the Swiss subgroup was healthier and more active than the full trial and started with lower omega-3 blood levels, which the authors suggest may have helped. Participants with lower starting omega-3 did show bigger shifts, but that subgroup split was exploratory. Fourth, there were only two blood draws, three years apart, and clock readings are noisy. Finally, one author disclosed founding a nonprofit that licenses epigenetic-clock patents, including GrimAge. The DNA assays themselves were run blinded to treatment.
Where other studies agree. A 2025 Mendelian randomization study, which uses inherited genetic variants as stand-ins for lifelong habits, found that genetically predicted oily-fish intake was linked to lower PhenoAge acceleration (p < 0.0086) and genetically predicted fish-oil supplement use to lower GrimAge (p = 0.037), both in the age-slowing direction. That is a genetic-proxy design, not a trial, and the p-values are modest. On the harder endpoint of death, a UK Biobank analysis of 117,702 people followed for 12.7 years found that the highest quintile of blood DHA had lower all-cause mortality than the lowest (HR 0.79, 95% CI 0.74 to 0.85). That is an association. People with high DHA levels tend to eat and live differently in many other ways, and the number cannot show that adding a capsule would give the same result.
Where they do not. VITAL, a trial of 25,871 middle-aged and older Americans, measured telomere length, a different aging marker, in 1,031 participants. One gram a day of marine omega-3 for 4 years had no significant effect on telomere length at year 2 or year 4. Vitamin D3 at 2,000 IU a day was the one that mattered there, reducing telomere shortening by 0.14 kb (95% CI 0.007 to 0.27). Telomeres and methylation clocks measure different things, so this is not a direct contradiction, but it shows how aging markers can disagree, and which supplement "worked" flipped between the two trials. An older 4-month trial of 106 overweight, sedentary middle-aged and older adults found no significant group difference in telomere length with 1.25 or 2.5 g of omega-3 a day. The F2-isoprostanes it used to track oxidative stress were about 15% lower in the supplemented groups, and telomere length did rise as the omega-6 to omega-3 ratio fell (p = 0.02).
What DO-HEALTH found on health outcomes. The trial's main report in JAMA in 2020 covered all 2,157 participants and found no statistically significant benefit from omega-3, vitamin D or the exercise program on any of six primary outcomes: systolic and diastolic blood pressure, physical performance, cognition, nonvertebral fractures and infections. For omega-3, systolic blood pressure changed by -0.8 mm Hg (99% CI -2.1 to 0.5), and the infection rate ratio was 0.89 (99% CI 0.78-1.01), close to, but short of, the trial's stricter significance bar. Only 25 participants died over the three years, spread evenly across groups, far too few to say anything about lifespan.
What to take from it. Read together, one good randomized trial sees a small shift in three clocks, two studies using other aging markers do not see an omega-3 effect, and the same trial did not show clear clinical gains. That makes this a hypothesis worth testing in larger and longer trials, not a reason to change what you do. The omega-3 in the trial came from algae, so fish is not required, and oily fish about twice a week remains the sensible food-level default. Nothing here argues for megadosing either: the trial used 1 g a day, and the same trial found no clear clinical gains. A mail-order biological-age test cannot tell you whether omega-3 is working for you either.
Bottom line. Can omega-3 slow biological aging? Maybe a little, on certain clocks, in healthy older adults, over three years. Whether that translates into living longer or staying healthier is still unknown, and the honest summary is that the result is interesting, small and unproven where it counts.
Medical note. This article is general education and not personal medical advice. Talk with a clinician before starting or changing a supplement, especially if you take blood-thinning medication or have a heart-rhythm condition.