Omega-3 (EPA/DHA)
If you eat fatty fish twice a week you probably don't need a supplement. If you don't, omega-3 is one of the few supplement categories with both a real (if modest) cardiovascular signal and a newer trial hint at slower biological aging — and the form, dose, and freshness of the product matter more than the brand.
EPA (eicosapentaenoic acid) and DHA (docosahexaenoic acid) are the two long-chain marine omega-3 fatty acids that account for the cardiovascular, brain-health, and now biological-aging signals in this category. ALA (alpha-linolenic acid) — the plant omega-3 in flaxseed, chia, and walnuts — converts to EPA/DHA poorly in humans (typically under 10%, with negligible yield of DHA) and is not a reliable substitute. The most useful framing: if your dietary EPA+DHA intake is adequate, supplementation is unnecessary; if it isn't, supplementation closes a measurable physiological gap that can be tracked on a single blood test.
What the evidence says
Weak-to-Moderate — for preventing heart attacks and cardiovascular death. Broad low-dose supplementation is largely neutral (Cochrane, VITAL, ASCEND); the one clearly positive high-dose trial (REDUCE-IT, EPA-only) is partly confounded by its placebo, and the matched EPA+DHA trial (STRENGTH) was null.
Moderate (as association), Weak (as causal claim) — for lower all-cause mortality tied to a higher omega-3 index. The signal is consistent across large cohorts but observational; the randomized mortality data are essentially neutral.
Weak — for slower biological aging. One post-hoc trial readout (DO-HEALTH) slowed methylation "aging clocks" modestly, but these are surrogate markers, not hard outcomes.
Moderate — for lowering triglycerides. Reliable and dose-dependent, the best-established pharmacological effect in the category.
Caution — for atrial fibrillation and bleeding at high doses. A small excess of atrial fibrillation shows up in trials above about 1.5 g/day of EPA+DHA in people at high cardiovascular risk: about 40% higher odds, under 1 extra case per 100 people. At 1.5 g/day or less, the largest (industry co-funded) analysis found no clear excess.[1] An earlier pooling of 7 heart trials did see a slight rise even at 1 g/day or less (HR 1.12).[2]
Two parallel stories
The literature now runs on two parallel tracks that need to be read together:
- The cardiovascular trial track. Large randomized trials at gram-level doses, looking at heart attacks, strokes, and cardiovascular death. The results here are mixed and have a real safety signal at the top of the dose range.
- The biomarker / longevity track. Observational cohorts and a handful of newer trials anchored on the omega-3 index (the EPA+DHA percentage of red-blood-cell membranes), and on epigenetic age. The signal here is more uniformly positive and dose-responsive — but the outcomes are softer (all-cause mortality, biological-age clocks) than hard events.
Almost all of the practical guidance below sits at the intersection of the two: enough EPA+DHA to bring your omega-3 index to about 8%, while staying at or below about 1.5 g/day, above which the trial excess of atrial fibrillation is clearest.
The cardiovascular trial puzzle
Start with the baseline the big syntheses establish, because it anchors everything that follows. The largest one — [3], 86 RCTs and 162,796 participants — found that increasing long-chain omega-3 has little or no effect on all-cause mortality: a 3% difference well within the range expected from chance (relative risk 0.97, 95% confidence interval 0.93–1.01; high-certainty). That bracketed range is where the true effect most plausibly lies; because it crosses 1.0, the result is compatible with no effect. Cardiovascular events told the same story — about 4% lower, but still compatible with no effect (RR 0.96, 0.92–1.01) — though omega-3 reliably lowered triglycerides (by about 15%) and had little or no effect on HDL ("good") cholesterol. A more optimistic counterweight, [4] (13 RCTs, 127,477 participants), found small but real reductions — about 8% each — in myocardial infarction (MI, or heart attack; RR 0.92, 0.86–0.99) and coronary death (RR 0.92, 0.86–0.98), but not stroke, with a dose-response hinting at benefit above the ~840 mg/day used in most trials. So the honest starting point for hard endpoints is neutral-to-modest, and the positive longevity framing below has to be read against it. For the dietary side of this story — fish intake, fat type, and the seed-oil debate — see Dietary fats.
Against that baseline sits a puzzle at the top of the dose range. The trial that put high-dose EPA back on the map: [5]. Icosapent ethyl (purified EPA, 4 g/day) cut major cardiovascular events by 25% over five years on top of optimal statin therapy. The absolute risk reduction was clinically meaningful — about 4–5 events prevented per 100 high-risk patients treated.
But STRENGTH, a similar trial using a mixed EPA+DHA formulation (carboxylic-acid form, 4 g/day), was null for cardiovascular events.[6] Two big differences may explain the discrepancy:
- The molecule. Pure EPA versus an EPA+DHA mix. Pooling 16 trials, purified EPA was linked to about 21% fewer cardiovascular deaths (hazard ratio, HR, 0.79) and EPA+DHA mixes to about 8% fewer (HR 0.92). The gap between the two was not statistically significant, and most of the EPA trials were open-label studies from Japan without a placebo.[7] Some lab research suggests DHA may partly offset EPA's effects, but the case that the molecule matters still rests mostly on REDUCE-IT.
- The placebo. REDUCE-IT used mineral oil, which after a year had raised LDL ("bad") cholesterol by about 11% and high-sensitivity C-reactive protein (hsCRP, an inflammation marker) by about 22% in the control group — possibly flattering the active drug by comparison.[8] STRENGTH used corn oil, which is more biologically inert. The counterpoint: the US Food and Drug Administration's review estimated that mineral oil's effect on these markers could explain at most about 3% of the benefit,[9] while the authors of the REDUCE-IT biomarker substudy called its impact on the result uncertain.[10] The controversy is real but unresolved. A 2024 Japanese trial with no mineral-oil comparator (the control group simply got no extra treatment) gives a partial answer. In 2,506 statin-treated people with coronary disease and a low ratio of EPA to arachidonic acid (an omega-6 fat) in their blood, 1.8 g/day of purified EPA for 5 years was followed by about 21% fewer cardiovascular events (9.1% vs 12.6%, HR 0.79). That is similar in size to REDUCE-IT, but the result just missed statistical significance, and new atrial fibrillation roughly doubled (3.1% vs 1.6%).[11] Patients knew which treatment they were getting, so the trial supports REDUCE-IT without settling the question.
The lower-dose primary-prevention trials round out the pattern. [12] (25,871 adults, 840 mg/day) was null on its primary composite — an 8% difference compatible with chance (hazard ratio 0.92, 0.80–1.06); the roughly 28%-lower heart-attack rate (HR 0.72, 0.59–0.90) and the benefit in low-fish-intake participants were secondary/exploratory findings, not headline results. [13] (15,480 people with diabetes, 1 g/day) was null for serious vascular events (8.9% vs 9.2%). [14] (1,027 elderly post-MI patients, 1.8 g/day vs corn oil) was null and showed a non-significant rise in new-onset atrial fibrillation.
The honest read: low-dose supplementation in unselected populations is largely neutral for hard endpoints (Cochrane, ASCEND, VITAL primary); the one clearly positive high-dose trial (REDUCE-IT, EPA-only) is partly confounded by its placebo; mixed high-dose EPA+DHA (STRENGTH, OMEMI) is null. For a healthy midlife adult, eating oily fish regularly is the well-supported option. Heart guidelines do not back over-the-counter fish oil: the 2026 US cholesterol guideline rates dietary supplements as "no benefit" for lowering cholesterol or triglycerides in people with abnormal blood lipids. It notes that non-prescription fish oil has not shown clinical benefit for high triglycerides or heart disease, and lists a rise in LDL cholesterol and atrial fibrillation as possible harms.[15] If you don't eat fish and want a modest daily supplement to raise your omega-3 index, that is a personal choice, not a guideline recommendation. The only omega-3 with guideline support for preventing heart attacks is prescription icosapent ethyl (purified EPA, 4 g/day). It is only for people already on a statin who are at high cardiovascular risk (in the US guideline: age 50 or over, with cardiovascular disease or with diabetes plus another risk factor) and whose triglycerides stay raised (about 1.7–5.6 mmol/L in the US guideline, which also requires LDL below 2.6 mmol/L; 1.5–5.6 mmol/L in the European one). The US guideline says it "may be reasonable"[16] and the European guideline says it "should be considered".[17] That is a decision for your doctor.
The longevity and biological-aging signal
The newer and arguably more interesting evidence comes from biomarker cohorts and from epigenetic-age trials.
In long-running cohorts, people in the top fifth for blood omega-3 levels have about 15–18% lower all-cause mortality than those in the bottom fifth (17 cohorts, about 42,000 adults).[18] A 2024 analysis that added UK Biobank to these cohorts covers more than 160,000 people. In it, the highest blood DHA levels went with about 17% lower all-cause mortality (HR 0.83), and the link held for heart, cancer and other deaths.[19] These come from the same research group. The Framingham cohort, which did measure the omega-3 index, found that people in the top fifth (above 6.8%) had about a third lower all-cause mortality than those in the bottom fifth (below 4.2%).[20] The mortality signal is most consistent for cardiovascular death but extends to cancer and non-cardiovascular causes in pooled analyses. The dose-response is non-linear for EPA (largest gain in moving from deficient to sufficient, then plateaus) and more linear for DHA.
Two caveats keep this honest. These are observational associations, and fish-eaters are systematically healthier — residual confounding is the central limitation, and it cuts hard given that the randomized mortality data are essentially neutral (VITAL all-cause death HR 1.02; Cochrane little or no effect). And the omega-3 index itself is promoted heavily by a commercial testing company whose founder co-authors much of this literature, so the specific ">8% target" should be read as a cardioprotective goal, not a longevity threshold the hard-outcome data require.
The readout that elevated omega-3 from "cardiovascular nutrient" to "candidate longevity intervention" is the [21] (Bischoff-Ferrari et al., Nature Aging 2025). One gram per day of algal-derived omega-3 (330 mg EPA + 660 mg DHA) slowed several next-generation methylation clocks — PhenoAge, GrimAge2, and DunedinPACE — by the equivalent of roughly three to four months of biological age over three years. DunedinPACE, which estimates the current rate of aging rather than cumulative age, slowed by about 1%. The effect was additive with vitamin D and a light home-exercise program, though that additivity appeared only on PhenoAge, not GrimAge2 or DunedinPACE; first-generation clocks (Horvath, Hannum) were unaffected.
A few honest caveats — and they matter here:
- This is a post-hoc analysis of a subset (777 of 2,157 participants, all from the Swiss sites), and the main DO-HEALTH trial actually missed all six of its pre-registered primary endpoints (blood pressure, non-vertebral fractures, physical function, cognition, infection rate). The clock benefit is an exploratory readout, not a headline result, and the standardized effects are small (d ≈ 0.16–0.32). DunedinPACE slowed ~1%, versus 2–3% in the CALERIE caloric-restriction trial.
- The comparator was sunflower oil, not an inert placebo, and the cardiovascular readout was mixed. Against it, omega-3 raised HDL and lowered triglycerides but raised calculated LDL cholesterol by 0.11 mmol/L, and it did not reduce major cardiovascular events (hazard ratio 1.00, 95% CI 0.64–1.56) — an exploratory endpoint with only 81 events, too few to test it. DSM Nutritional Products, a supplement maker, was among the trial's funders.[22]
- The clocks are surrogate biomarkers of biological age, not hard outcomes. They predict mortality and disease in cohorts, but a slowed clock in a three-year trial is not the same as a longer life.
- The participants were over 70 — older than the typical longevity-curious 40-something — and the paper did not report their omega-3 index. Over a third (38%) started with low blood EPA+DHA (below 100 ng/mL), and the clock effect looked somewhat larger in people who started lower.[23]
- The clock-slowing dose (1 g/day of EPA+DHA) is well below the high-dose cardiovascular-trial range. The longevity signal does not require — and probably does not benefit from — pushing the dose to 4 g/day.
Mechanistically the case is coherent: EPA and DHA integrate into cell membranes, dampen the chronic low-grade inflammation of aging, shift lipid mediators away from pro-inflammatory prostaglandins toward specialized pro-resolving mediators (resolvins, protectins, maresins), and modulate the same nutrient-sensing and cellular-recycling pathways (the AMPK energy sensor, the SIRT1 and mTOR growth-and-repair switches, and autophagy) targeted by caloric restriction and rapamycin. None of that proves causality on its own, but it makes the clock data less surprising.
Test: the omega-3 index
The most useful single lab in this space is the omega-3 index — the percentage of EPA + DHA in red blood cell membranes:
| Omega-3 index | Status |
|---|---|
| <4% | Low; associated with elevated cardiovascular risk and higher all-cause mortality |
| 4–8% | Intermediate |
| 8–12% | Target range |
| >12% | High; reconsider dose, especially before surgery or with anticoagulants |
The 8% target comes mainly from cardiovascular research, not from a mortality threshold; the cohort signals above compare top and bottom fifths rather than testing cut-offs. From a five-month dosing trial in young adults who ate little fish, the researchers estimated that about 1.2 g/day of EPA+DHA would be enough to lift a 75 kg adult from a typical starting index of about 4.3% to 8%, with heavier people needing more; 1.8 g/day gave an average of about 9.5%.[24] People who eat fatty fish two to three times weekly often reach it without supplementation. Athletes and habitual non-fish-eaters routinely test in the 4–5% range — i.e. the bottom of the intermediate band — even when they consider their diets healthy. A cross-sectional analysis of collegiate athletes found a mean omega-3 index of 4.6%,[25] and the International Society of Sports Nutrition (ISSN) position stand reports a mean of 4.4% across 404 collegiate football players, not one of whom reached 8%.[26]
The omega-6 to omega-3 ratio
A second framing worth knowing, even though it's less actionable than the omega-3 index itself: the ratio of plasma omega-6 to omega-3 polyunsaturated fats. In a UK Biobank analysis of around 85,000 adults followed for an average of 13 years, the highest-ratio quintile carried roughly 26% higher all-cause mortality, 31% higher cardiovascular mortality, and 14% higher cancer mortality than the lowest-ratio quintile.[27] Modern Western diets frequently sit at 15:1 or higher, against an evolutionary baseline closer to 1:1 to 5:1.
The practical implication is not to chase a target ratio — that's hard to measure cleanly outside of research labs, and the absolute omega-3 level seems to do most of the work. It is to recognize that raising EPA+DHA is the easier lever than cutting omega-6: seed oils, processed snack foods, and most restaurant cooking deliver large omega-6 loads passively. Adding fish or a supplement shifts the ratio in the right direction without requiring a dietary overhaul.
Brain, mood, and cognition
DHA is the dominant structural fatty acid in the brain; EPA is the active mood-and-inflammation lever. The clinical literature reflects that split.
In broad trials of cognitively healthy older adults, omega-3 supplementation has not moved global cognition reliably. Pooled trial data point to at most a small benefit. A 2025 overview combining 14 randomized trials in about 27,000 adults aged 40 and over (without dementia, some with mild cognitive impairment) found a slight improvement on a standard memory-and-thinking screening test (effect size 0.16, with the lower end of the plausible range barely above zero). Higher doses or longer use did not help more.[28] A 2025 dose-response analysis of 58 trials reported a larger overall gain that peaked at about 1.5 g/day, but its certainty was rated low, the trials disagreed sharply and there were signs of publication bias. In that analysis's subgroups of cognitively healthy adults and of adults over 60, the effect was not significant.[29] The trials are heterogeneous (doses, EPA:DHA ratios, durations, baseline status), and the broad-population effect is consistent with a small, hard-to-detect signal in the average healthy adult.
In an observational cohort of 323 cognitively unimpaired middle-aged adults at high risk of Alzheimer's (the ALFA study), higher red-blood-cell DHA went with slightly better scores on a sensitive memory-and-thinking composite over three years; the link did not differ between carriers and non-carriers of APOE-ε4, the most common inherited risk gene for Alzheimer's.[30] Whether people at higher genetic risk benefit from supplements is unproven. In a two-year randomized trial, 365 adults aged 55–80 with low DHA intake (nearly half carrying APOE-ε4) took 2 g of DHA a day or placebo. The DHA reached the brain, but thinking and brain volume did not differ from placebo; the trial was small for those outcomes and lost 38% of participants.[31] A three-year trial in 102 adults aged 75 and over also missed its main brain-imaging goal, with only a hint of benefit on a secondary measure in APOE-ε4 carriers.[32]
For depression, omega-3 has a small, uncertain effect. Pooled across 36 randomized trials, it lowered symptom scores slightly (standardized difference -0.26), but it did not raise the share of patients who responded or recovered, and the trials varied widely and showed signs of publication bias.[33] In that analysis, the best results came from 1–1.5 g/day at an EPA:DHA ratio between 1:1 and 2:1, taken for at least 8 weeks, in Asian patients with mild-to-moderate depression who were not on other medication. The International Society for Nutritional Psychiatry Research guideline for major depressive disorder favours EPA-rich products instead: pure EPA, or an EPA:DHA ratio above 2:1, at 1–2 g of EPA per day. It also recommends that a clinician confirm the diagnosis first and monitor for side effects.[34] No meta-analysis has settled on a single required ratio.
Whether inflammation picks out the patients who benefit is still an open question. Small trials hint that high-dose EPA may help people with depression and raised inflammation markers: in one trial, 61 unmedicated, overweight adults with depression and raised C-reactive protein (CRP, an inflammation marker) were split between three EPA doses and placebo. Among those who finished, 64% responded on 4 g of EPA a day versus 40% on placebo, a difference that was not statistically significant.[35] A 2025 review of recent trials calls high-dose EPA in inflammatory depression promising but says the right dose and the right patients are not yet established.[36] If you try omega-3 for low mood, the specialist guideline favours an EPA-rich product, though the pooled trials do not clearly favour one ratio. It should go alongside treatment, not replace it.
Taking fish oil to prevent depression is not supported. In the largest prevention trial, 18,353 adults aged 50 and over without depression symptoms took 1 g of fish oil or placebo for about five years. Mood scores did not differ, and the risk of depression was slightly higher in the fish-oil group (HR 1.13, about 13% higher).[37] Even as a treatment in older adults, the benefit is small: a 2025 meta-analysis found omega-3 lowered depressive symptoms only slightly (standardized difference -0.14), across trials that varied widely.[38]
Practical dosing
- Two or more servings of fatty fish per week (salmon, sardines, mackerel, herring, trout), ideally not only farmed salmon — supplementation is unnecessary.
- No fish in your diet: 1–1.5 g/day combined EPA+DHA. This is the band that lifts most people's omega-3 index to about 8% and starts at the DO-HEALTH clock-slowing dose of 1 g/day. It stays at or below the 1.5 g/day level above which the trial excess of atrial fibrillation is clearest, though it is not completely free of rhythm risk (see The atrial fibrillation paradox). If you have atrial fibrillation or are at high risk of it, agree the dose with your doctor.
- Athletes and heavy training loads: the same 1–1.5 g/day band, aimed at an omega-3 index of about 8%. The 2025 ISSN position stand concluded that athletes are at elevated risk of omega-3 insufficiency, that supplementation may attenuate indirect markers of muscle damage and subjective soreness, and that it can favourably shift immune-cell and cytokine responses (a reproducible drop in tumour necrosis factor-α (TNF-α), an inflammatory signalling protein, at roughly 2.4 g EPA + 1.2 g DHA per day over four weeks or more). It stops short of naming a single dose — trials span 0.9–6.3 g/day — but estimates that about 1.4 g/day of extra EPA+DHA is what it takes to lift a typical athlete's omega-3 index to 8%.[39]
- Elevated triglycerides or established cardiovascular disease: higher doses (2–4 g/day combined, or — with a prescriber — icosapent ethyl 4 g/day) may be appropriate. The [40] concluded prescription omega-3 at 4 g/day lowers triglycerides by 20–30%, scaling with baseline level.
- Vegetarians and vegans: algae-derived oils provide DHA and increasingly EPA. In one industry-funded randomized trial (74 adults, 14 weeks), algal oil was non-inferior to fish oil for raising EPA+DHA in blood plasma, after allowing for the different doses; red-cell levels were not measured.[41] Microalgae are also the original biological source — fish accumulate EPA and DHA only by eating algae.
Regulatory ceilings are broadly aligned: the European Food Safety Authority concluded long-term combined EPA+DHA up to 5 g/day is safe for healthy adults, and the US Food and Drug Administration (FDA) concluded in 2019 that supplements providing up to 5 g/day of EPA+DHA, as labelled, are safe; foods with added omega-3 are held to a lower 3 g/day total.[42]
Form, sourcing, and oxidation
The product-quality question is, for most consumers, where omega-3 supplementation actually goes wrong. Three issues stack: the chemical form, the source, and oxidation.
Form
| Form | Notes |
|---|---|
| Triglyceride / re-esterified triglyceride (rTG) | The natural form; best absorption, especially without a high-fat meal. Roughly 30% higher EPA/DHA uptake than ethyl ester in head-to-head pharmacokinetic studies. Premium price. |
| Ethyl ester (EE) | The most common pharmaceutical form (icosapent ethyl, omega-3 acid ethyl esters). Requires pancreatic lipase to cleave the ethanol; absorption gap narrows when taken with a fatty meal. Acceptable; cheaper. |
| Phospholipid (krill oil) | Marginally better absorption per gram, but much lower EPA+DHA content per capsule. Expensive per gram of EPA+DHA delivered. |
| Free fatty acid | Used in STRENGTH (Epanova); withdrawn after the trial failed. |
For most healthy adults, a third-party-tested rTG or ethyl ester product taken with food is adequate. The cost difference between rTG and ethyl ester is usually 2–3×; the absorption difference, with meals, is small.
Sourcing
Wild-caught small forage fish (sardines, anchovies, herring, mackerel) remain the highest-density, lowest-contaminant dietary source. Farmed Atlantic salmon used to be a comparable source, but its EPA+DHA content has fallen as fish-farm feeds switched from fish oil to cheaper plant oils. Eight farmed salmon fillets bought in US shops around Washington DC in 2023 contained about 0.9 g EPA+DHA per 100 g raw fillet. That is less than half the 2.0 g/100 g still listed in older USDA tables, and the fillets had more than twice as much omega-6 linoleic acid.[43] At that level, two 150 g portions a week supply only about 0.4 g/day, so people whose only oily fish is farmed salmon may still fall short of an 8% omega-3 index. Algal oil supplements deliver the same EPA and DHA molecules and bypass both the contaminant and sustainability concerns; one 2025 randomized trial found algal oil as good as fish oil, dose for dose, for raising plasma omega-3 levels.[44]
Oxidation and freshness
The same carbon-carbon double bonds that make EPA and DHA biologically active also make them chemically fragile. Heat, light, and ambient oxygen during manufacture, transport, and storage produce peroxides and secondary aldehydes that the industry summarizes as the TOTOX (total oxidation) value. An independent analysis of 171 North American over-the-counter omega-3 products found that half of the products tested exceeded at least one voluntary industry oxidation limit, with flavoured, bulk and children's products the worst.[45] A later US analysis of 72 products tested between 2014 and 2020 found the problem mostly in flavoured products: about two in three flavoured supplements (68%) went over the industry limit for total oxidation, set by the omega-3 trade body GOED, against about one in eight unflavoured ones (13%).[46]
Oxidised omega-3 seems mainly to lose its benefit. In a small 7-week randomized trial of 54 healthy adults taking 1.6 g/day EPA+DHA, fresh fish oil lowered LDL and related cholesterol-carrying particles more than both a deliberately oxidised fish oil and a sunflower-oil control, so the oxidised oil lost that benefit.[47] Trials have not shown that rancid oil actively harms people, but there is no point paying for a product that may not work.
Practical filters:
- Look for an IFOS (International Fish Oil Standards), Labdoor, USP (U.S. Pharmacopeia), or NSF mark on the bottle. IFOS specifically publishes TOTOX numbers and heavy-metal panels per batch.
- Check for a freshness signal on the certificate of analysis (TOTOX, peroxide value, anisidine value).
- Refrigerate after opening and finish the bottle within a few months.
- Trust your nose. A high-quality oil smells nearly neutral. Strong fishy odour or post-dose fishy reflux is the sensory signature of advanced oxidation, not of "real" fish oil.
The atrial fibrillation paradox
This deserves its own section because two genuinely contradictory-looking lines of evidence coexist.
The trial signal. Several high-dose randomized trials of purified or synthetic omega-3 (mostly ethyl ester, mostly at 1–4 g/day, in older patients with established cardiovascular disease) showed a small but real increase in new-onset atrial fibrillation. A pooled analysis of 7 cardiovascular-outcome trials in 81,210 patients found about a 25% higher rate of new atrial fibrillation versus placebo (hazard ratio 1.25, 95% CI 1.07–1.46), and the excess was clearly dose-dependent — roughly 49% higher above 1 g/day (HR 1.49, 1.04–2.15) versus about 12% higher at ≤1 g/day (HR 1.12, 1.03–1.22), with each additional gram per day adding ~11% (HR 1.11).[48] Both REDUCE-IT and STRENGTH contributed at the 4 g/day dose.
The biomarker signal — pointing the other way. When the question is asked using physiological status rather than pharmacological dose, the relationship flips. In the UK Biobank, people with higher blood omega-3 levels had less new atrial fibrillation: about 11% less per interquintile-range step up in level (HR 0.89), and about 14% less in the top fifth than in the bottom fifth (7.5% vs 6.9% over about 13 years).[49] Whether taking fish oil does the same for healthy people is disputed. One analysis of the same cohort found that regular fish-oil users without heart disease had about 13% more new atrial fibrillation (HR 1.13) and slightly more stroke (HR 1.05, borderline).[50] A later analysis of the same cohort, by researchers with commercial ties to omega-3 products, found no link between fish-oil use and atrial fibrillation (HR 1.00) once age was modelled in finer detail; a small link remained only when age was split into over/under 65.[51] Both are observational, so neither settles the question.
A reasonable reading: the largest trial analysis (35 trials, co-funded by an omega-3 industry body) found no clear rise in atrial fibrillation at 1.5 g/day of EPA+DHA or less, the range most supplement users take. Above 1.5 g/day in people already at high cardiovascular risk, the odds rose by about 40% (OR 1.43), about 1 extra case per 125 people treated. Pooled across all trials, the odds were still about 14% higher.[52] The lower risk seen with higher blood omega-3 levels is observational and does not show that supplementing protects you. At 4 g/day, the extra atrial fibrillation is a trade-off worth making only where the high dose has shown benefit, namely EPA-only icosapent ethyl in high-risk patients.
Cautions
- Atrial fibrillation signal at high doses (Caution). As above: in trials the excess is clearest above about 1.5 g/day in people at high cardiovascular risk; an older pooling of heart trials saw a slight rise even at 1 g/day or less (HR 1.12).
- Bleeding. In a pooled analysis of large randomized trials with over 120,000 participants (funded by the maker of icosapent ethyl), mostly older adults with or at risk of heart disease, omega-3 did not significantly raise overall bleeding, bleeding strokes or gut bleeding (any bleeding 2.6% vs 2.4%, RR 1.09).[53] The exception was prescription high-dose purified EPA (icosapent ethyl, 4 g/day), which raised bleeding by about half in relative terms, roughly 1 extra bleed for every 166 people treated, and no clear rise in serious bleeding.[54] Stopping high doses about a week before elective surgery is a common precaution, but it rests on caution rather than trial evidence.
- Drug interactions. Across trials, extra bleeding tracked the EPA dose rather than how many participants were also on antiplatelet drugs such as aspirin. This was a trial-level comparison, not a test in individual aspirin users.[55] That is mildly reassuring for ordinary doses alongside aspirin. People on blood thinners such as warfarin or apixaban should still check with their prescriber before taking high doses, because those drugs were not analysed separately.
- Pregnancy / lactation. DHA needs rise in pregnancy. A 2024 international guideline advises at least 250 mg/day EPA+DHA from food or supplements for women of childbearing age, plus an extra 100–200 mg/day of DHA during pregnancy. Pregnant women with a low DHA intake or low blood DHA should take about 600–1,000 mg/day of EPA+DHA (or DHA alone) to lower the risk of preterm birth. They should start in the second trimester, no later than about 20 weeks, and continue until about 37 weeks or birth.[56] Pooled trials show about 17% lower odds of preterm birth with supplementation (OR 0.83).[57] Choose products with verified low contaminant panels.
What to skip
- ALA-only supplements (flaxseed oil, chia oil) — ALA converts to EPA/DHA at single-digit-percent rates and yields negligible DHA. Useful as food, not as a substitute for marine omega-3.
- Generic "fish oil" without a third-party oxidation certificate. Rancid product is common, especially on unrefrigerated retail shelves, and undoes most of the point of supplementing.
- Krill oil at premium prices — the absorption advantage doesn't justify the per-gram cost of EPA+DHA delivered for most people.
- Flavoured, gummy, or chewable omega-3 for adults — the freshness data on these formats is consistently the worst, and the per-capsule EPA+DHA load is low.
Further reading
- Bhatt DL et al. Cardiovascular Risk Reduction with Icosapent Ethyl for Hypertriglyceridemia (REDUCE-IT). NEJM 2019.[58]
- Nicholls SJ et al. Effect of High-Dose Omega-3 Fatty Acids vs Corn Oil on Major Adverse Cardiovascular Events (STRENGTH). JAMA 2020.[59]
- Manson JE et al. Marine n-3 Fatty Acids and Prevention of Cardiovascular Disease and Cancer (VITAL). NEJM 2019.[60]
- Harris WS et al. Blood n-3 fatty acid levels and total and cause-specific mortality from 17 prospective studies (FORCE consortium). Nature Communications 2021.[61]
- Bischoff-Ferrari HA et al. Individual and additive effects of vitamin D, omega-3 and exercise on DNA methylation clocks (DO-HEALTH). Nature Aging 2025.[62]
- Abdelhamid AS et al. Omega-3 fatty acids for the primary and secondary prevention of cardiovascular disease (Cochrane). Cochrane Database Syst Rev 2020.[63]
- ASCEND Study Collaborative Group. Effects of n-3 Fatty Acid Supplements in Diabetes Mellitus. NEJM 2018.[64]
- Gencer B et al. Effect of Long-Term Marine n-3 Fatty Acids Supplementation on the Risk of Atrial Fibrillation in RCTs of Cardiovascular Outcomes. Circulation 2021.[65]
- Skulas-Ray AC et al. Omega-3 Fatty Acids for the Management of Hypertriglyceridemia: A Science Advisory From the AHA. Circulation 2019.[66]
- O'Keefe E et al. Associations Between Plasma Omega-3 and Fish Oil Use With Risk of Atrial Fibrillation in the UK Biobank. J Am Heart Assoc 2025.[67]
- Jackowski SA et al. Oxidation levels of North American over-the-counter n-3 (omega-3) supplements and the influence of supplement formulation and delivery form on evaluating oxidative safety. J Nutr Sci 2015.[68]
- Jäger R et al. International Society of Sports Nutrition Position Stand: Long-Chain Omega-3 Polyunsaturated Fatty Acids. J Int Soc Sports Nutr 2025.[69]
- Lázaro I et al. Red blood cell ω-3 status and longitudinal cognition in individuals at risk of Alzheimer disease (ALFA+). J Nutr 2025.[70]
- Heileson JL et al. A Cross-Sectional Analysis of Whole Blood Long-Chain ω-3 Polyunsaturated Fatty Acids and Its Relationship with Dietary Intake, Body Composition, and Measures of Strength and Power in Collegiate Athletes. J Am Nutr Assoc 2023.[71]