Fish Oil and Atrial Fibrillation: What the Heart Trials Found

Fish oil spent decades with a reputation as a friend to the heart's rhythm. Laboratory work and early trials suggested that the omega-3 fatty acids EPA and DHA might calm irritable heart cells, and that idea found its way into countless supplement labels and articles. Then the large heart trials of the past decade reported something few people expected: at high doses, people taking omega-3 developed atrial fibrillation more often, not less.

Atrial fibrillation, usually shortened to AF or AFib, is the most common sustained heart rhythm disorder. The upper chambers of the heart quiver instead of contracting in a steady rhythm, which can cause palpitations, breathlessness and fatigue, and it raises the risk of stroke because blood can pool and clot inside the heart. This review walks through what each major trial found, where the signal shows up and where it does not, what regulators and cardiology guidelines now say, and what it means for the dose on your supplement label.

Why fish oil and AF became a question

The idea that fish oil protects the heart's rhythm grew out of the 1990s. In the Italian GISSI-Prevenzione trial, 11,324 people who had recently had a heart attack were assigned to 1 gram a day of omega-3, vitamin E, both, or neither. The omega-3 groups had fewer deaths (a 14 percent relative reduction in the main analysis), and together with laboratory work on heart cells, that result fed the theory that omega-3 steadies the heart's electrical system. But the trial was open-label, with no placebo capsule, so participants and doctors knew who was taking what.

When the theory was tested with placebo capsules in people with implanted defibrillators, it did not hold up well. In a US trial of 200 such patients, the main outcome, time to a first defibrillator treatment for a dangerous ventricular rhythm, showed no significant difference, but recurrent dangerous ventricular rhythms were more common on fish oil, and the authors warned it might trigger dangerous heart rhythms in some patients. A second trial of 402 patients found only a trend toward fewer defibrillator-treated rhythm events or deaths that did not reach statistical significance, and a European trial of 546 patients found no significant difference. Pooled, the three trials showed no convincing protection. A 2024 review in Circulation summed up the sudden-death findings as not widely replicated.

Omega-3s clearly do something to the heart's electrical behavior. A meta-analysis, a study that pools the results of earlier trials, combined 30 randomized trials and found that fish oil lowers the resting heart rate by about 1.6 beats per minute, more in people whose heart rate starts higher. The open question was whether those effects would help or hurt the upper chambers, where atrial fibrillation begins. The big cardiovascular outcome trials, designed to test heart attacks and strokes, ended up answering it.

A note on the numbers below. A hazard ratio (HR) compares the rate of an event between two groups, so 1.0 means no difference and 1.25 means a 25 percent higher relative rate. Odds ratios, rate ratios and relative risks are read the same way, with 1.0 meaning no difference. The 95 percent confidence interval (CI) is the range of values the data are compatible with; if it includes 1.0, the result is not statistically significant. A P value below 0.05 is the usual cutoff for calling a result statistically significant.

What the big heart trials found

These are the randomized trials that reported AF, from the highest dose to the lowest. The doses are the daily amounts of the product, and all participants were adults. The high-dose trials mostly used prescription drugs; the 1 gram trials used capsules similar in strength to many concentrated supplements.

Summaries describe research in adults, mostly with heart disease, diabetes or cardiovascular risk factors. AF was defined differently in each trial, and only VITAL Rhythm made it the main outcome. Nothing here is a claim that any supplement prevents or treats a disease.

The high-dose trials

REDUCE-IT. This trial randomized 8,179 statin users who had raised triglycerides and either established heart disease or diabetes plus other risk factors. They were assigned to 4 grams a day of icosapent ethyl, sold as Vascepa, or a mineral oil placebo for a median of 4.9 years. Icosapent ethyl is a highly purified prescription form of EPA. Its main composite of heart attacks, strokes, cardiovascular deaths and related events fell from 22.0 to 17.2 percent (HR 0.75). But hospitalization for AF or atrial flutter rose from 2.1 to 3.1 percent (P=0.004): 127 people on the drug versus 84 on placebo, a hazard ratio of 1.5 according to the FDA label. The trial was funded by Amarin, the drug's maker.

A later analysis, also funded by Amarin, split the result by AF history. Among the 751 people who already had AF, 12.5 percent on the drug were hospitalized for it versus 6.3 percent on placebo. Among the 7,428 without prior AF, the figures were 2.2 versus 1.6 percent, a difference that was not statistically significant on its own (P=0.09). The paper concludes that most AF hospitalizations occurred in people with a history of AF, but its own patient counts appear to point the other way: of the 211 patients hospitalized for AF, 70 were in that group and 141 were in people without prior AF. The rate was higher with prior AF; the number of patients was not. Only 3 patients had the study drug withdrawn because of an AF event.

Two further details matter. First, more AF did not translate into more stroke: fatal or nonfatal stroke occurred in 2.4 percent on the drug versus 3.3 percent on placebo (HR 0.72, 95% CI 0.55 to 0.93), according to the FDA label. Second, the placebo is debated. The 2023 US guideline for chronic coronary disease notes that mineral oil worsened lipid and inflammatory markers and may not be an inert placebo, which could have made the drug look better on heart outcomes. That debate concerns the benefit, though, not the AF result, and AF also rose in trials that used corn oil.

STRENGTH. This trial gave 13,078 high-risk statin users 4 grams a day of an omega-3 carboxylic acid drug containing both EPA and DHA, or corn oil. It was stopped early when an interim analysis showed little chance of benefit: the main heart outcome occurred in 12.0 percent versus 12.2 percent (HR 0.99). New-onset AF reported by the investigators occurred in 2.2 percent on omega-3 versus 1.3 percent on corn oil (HR 1.69, 95% CI 1.29 to 2.21), which works out to one extra case for every 114 people treated. The main paper's results table labels AF as not adjudicated, meaning it was not confirmed by an independent committee, although a 2026 follow-up analysis describes committee adjudication; the two papers do not agree. Nonfatal stroke showed no significant difference (2.2 versus 1.9 percent, HR 1.14, 95% CI 0.90 to 1.45). Gastrointestinal side effects were more common on omega-3 (24.7 versus 14.7 percent). In the 2026 analysis of people without prior AF, the ones who developed it on omega-3 tended to be older, male, heavier, and to have heart failure, insulin-treated diabetes or elevated C-reactive protein, a marker of inflammation. The trial was funded by AstraZeneca.

RESPECT-EPA. This Japanese trial assigned 2,506 people with stable coronary disease on statins to 1.8 grams a day of icosapent ethyl or to no added treatment, for a median of 5 years. The main heart outcome was lower on EPA (9.1 versus 12.6 percent) but narrowly missed statistical significance (HR 0.79, 95% CI 0.62 to 1.00, P=0.055). New-onset AF occurred in 3.1 percent on EPA versus 1.6 percent without it (P=0.017). Because the trial was open-label, with no placebo, AF reported as an adverse event could have been picked up more readily in the treated group, but it is still notable as a signal at 1.8 grams rather than 4.

OMEMI. This Norwegian trial gave 1,027 adults aged 70 to 82, two to eight weeks after a heart attack, 1.8 grams a day of an omega-3 product (930 mg EPA plus 660 mg DHA) or corn oil for two years. The main heart outcome showed no significant difference (21.4 versus 20.0 percent). New AF, confirmed by an independent committee blinded to treatment and analyzed in the 759 people without prior AF, occurred in 7.2 percent on omega-3 versus 4.0 percent on corn oil (HR 1.84, 95% CI 0.98 to 3.45, P=0.06), close to but not reaching statistical significance. A later analysis that added brief episodes caught on handheld ECG screening, which it called micro-AF and which is not a standard diagnosis, found 11.9 versus 6.5 percent (HR 1.90, 95% CI 1.16 to 3.11). One complication: participants were allowed a child's spoon of cod liver oil a day, about 600 mg of EPA plus DHA, and about one in five used it, evenly across both groups, so the placebo group was not omega-3 free and total intake in both groups was higher than the capsules alone suggest.

At supplement doses: VITAL, ASCEND and the rest

VITAL Rhythm. VITAL randomized 25,871 US men aged 50 and older and women aged 55 and older, without heart disease or cancer, to one capsule a day providing 460 mg EPA and 380 mg DHA, or an olive oil placebo, alongside a separate vitamin D comparison. The main trial found no significant reduction in major cardiovascular events (HR 0.92, 95% CI 0.80 to 1.06). Its AF study, VITAL Rhythm, excluded the 752 people who already had AF and followed the remaining 25,119 for a median of 5.3 years, with every case confirmed from medical records by a committee. AF occurred in 469 people on omega-3 (3.7 percent) and 431 on placebo (3.4 percent), a hazard ratio of 1.09 (95% CI 0.96 to 1.24, P=0.19): no significant difference. Analyses restricted to cases confirmed by ECG, or to people who kept taking their capsules, gave similar, nonsignificant hazard ratios of 1.12 to 1.13. This is the largest trial on the question and the only large one designed with AF as its primary outcome. It still has limits: brief or silent AF was probably missed, the participants were healthier than in the high-dose trials, and the upper end of its confidence interval does not rule out a modest increase.

ASCEND. This UK trial gave 15,480 adults with diabetes and no known heart disease a 1 gram capsule with 840 mg EPA plus DHA, or olive oil, for an average of 7.4 years, and found no significant effect on serious vascular events (8.9 versus 9.2 percent). Its main report counted AF from participant reports, which gave a rate ratio of 1.23 (95% CI 0.98 to 1.54) as tabulated in a 2021 meta-analysis. A 2020 research letter then added linked hospital records; with records and participant reports combined, AF was found in 1,177 participants, compared with 287 by self-report alone. With those records, AF occurred in 7.7 percent on omega-3 and 7.6 percent on placebo (rate ratio 1.02, 95% CI 0.91 to 1.15). People taking anticoagulants were excluded, which kept out most people with known AF. The same letter reported nonfatal ventricular rhythm problems in 81 versus 54 people (rate ratio 1.49, P=0.02) and fewer cardiac deaths on omega-3 (1.7 versus 2.2 percent, P=0.04); neither counted as statistically significant under the stricter P value threshold of 0.01 that the letter set for this analysis. Most of those ventricular events (106 of 135) required resuscitation, ablation or an implanted defibrillator, and restricting the analysis to them gave a rate ratio of 1.90 (95% CI 1.30 to 2.77). The authors called for trials to report arrhythmias systematically.

The ASCEND authors added an important caveat. They calculated that REDUCE-IT's excess corresponds to about a 10 percent increase per gram, and wrote that the association in the combined low-dose trials, while not statistically significant, does not exclude the level of effect per gram seen in REDUCE-IT. In other words, the 1 gram trials are too small to prove the absence of a small effect.

The other 1 gram trials. In GISSI-HF, among 5,835 people with chronic heart failure and no AF at the start, new AF occurred in 15.2 percent on 1 gram a day of omega-3 versus 14.0 percent on placebo (HR 1.10, P=0.19). In the Italian Risk and Prevention Study of 12,505 adults with several cardiovascular risk factors, AF listed as a reason for heart-related hospitalization, which was not independently confirmed, occurred in 1.8 versus 1.5 percent (HR 1.22, 95% CI 0.93 to 1.61), as tabulated in the 2021 meta-analysis. None of the 1 gram trials was significant on its own, but most of their point estimates sit slightly above 1.0, which is why the pooled analyses below matter.

What the pooled analyses say

The 2021 meta-analysis. Researchers from the VITAL team pooled seven cardiovascular outcome trials covering 81,210 people followed for an average of 4.9 years. Omega-3 was associated with a 25 percent higher relative risk of AF (HR 1.25, 95% CI 1.07 to 1.46). The increase was larger in trials above 1 gram a day (HR 1.49) than at or below 1 gram (HR 1.12, 95% CI 1.03 to 1.22), and the risk rose by about 11 percent for each additional gram. The low-dose result is fragile. When the authors swapped in ASCEND's hospital-record data, the low-dose estimate became 1.08 (95% CI 0.99 to 1.17, P=0.077) under their main statistical model, no longer significant, while a fixed-effect model gave 1.075 (95% CI 1.003 to 1.153, P=0.042), which they called marginally significant. Note that 1 gram here means capsule weight: those capsules held about 840 to 880 mg of EPA plus DHA.

Other pooled analyses. A 2021 meta-analysis of eight trials and 83,112 people found a similar picture: AF in 4.0 versus 3.3 percent (relative risk 1.24, 95% CI 1.11 to 1.38), with 1.12 at 1 gram or less and 1.51 above 1 gram, and no significant difference in stroke (relative risk 1.04, 95% CI 0.90 to 1.20). A smaller 2021 analysis of five trials reported a 37 percent higher incidence (rate ratio 1.37).

The largest analysis, from 2026. This meta-analysis gathered 35 randomized trials (37 data sets) covering 114,592 people, including previously unpublished AF data obtained from trial investigators. It split the trials by cardiovascular risk and by dose, at 1,500 mg a day of EPA plus DHA. Only one group showed a statistically significant increase: high-risk people on high doses (odds ratio 1.43, 95% CI 1.14 to 1.79), an absolute difference of 0.8 percentage points. The other three groups, high risk with low dose, low risk with low dose, and low risk with high dose, showed no significant difference (odds ratios 1.07, 1.06 and 1.03). The authors suggest the signal is tied to dose and background cardiovascular risk rather than to any one formulation. Two cautions apply. Twenty-five trial investigators did not respond to requests for data, so some trials are missing, and the low-risk, high-dose group had very few AF events. And the work was partly funded by GOED, the omega-3 industry's trade organization, which the paper says had no role in the analysis; its first author received a GOED research award, one author is chief medical officer of a company that sells omega-3 supplements, and another holds stock in a laboratory that sells omega-3 blood tests.

Reading them together. Every pooled analysis points the same direction, and those that split the trials by dose found a larger effect at higher doses, although in the 2026 analysis the increase was significant only in high-risk people on high doses. The disagreements are about the low end: whether about 1 gram a day carries a small increase or none. The absolute differences stay modest: about 1 to 1.5 percentage points over several years in the high-dose trials with significant results, and 0.8 percentage points in the 2026 pooled estimate for high-risk people on high doses, though larger in people who already had AF (12.5 versus 6.3 percent hospitalized for AF in REDUCE-IT).

Can fish oil prevent AF instead?

Before the safety signal emerged, researchers spent years testing the opposite idea: that fish oil could prevent AF. The results followed a pattern that shows up often in supplement research. Small open-label trials, with no placebo, looked impressive, and larger placebo-controlled trials did not confirm them.

After heart surgery. AF is common after heart surgery. A 2005 trial of 160 bypass patients, open-label with no placebo, reported AF in 15.2 percent of those given omega-3 versus 33.3 percent of controls. The 2012 OPERA trial then tested the idea properly, in 1,516 cardiac surgery patients in a multinational trial with an olive oil placebo: AF occurred in 30.0 percent on fish oil and 30.7 percent on placebo (odds ratio 0.96, 95% CI 0.77 to 1.20). A 2013 meta-analysis made the pattern explicit. Benefit appeared only in the open-label trials (odds ratio 0.40), while the placebo-controlled trials together showed no significant effect (odds ratio 0.92, 95% CI 0.78 to 1.10), and the difference between the two sets of trials was statistically significant.

In people who already have AF. An open-label trial of 178 people with persistent AF (AF that continues for more than seven days and needs treatment to stop) gave 6 grams a day of fish oil, starting more than a month before cardioversion, the procedure that shocks the heart back into normal rhythm, and continuing for up to a year. It reported recurrence at 90 days in 38.5 percent versus 77.5 percent of controls. The larger placebo-controlled trials told a different story:

• Prescription omega-3, 663 people. 8 grams a day for a week, then 4 grams, for 24 weeks. Symptomatic recurrence was 52 versus 48 percent in paroxysmal AF (AF that comes and goes, with episodes ending within seven days) and 50 versus 33 percent in persistent AF; combined, the hazard ratio was 1.22 (95% CI 0.98 to 1.52, P=0.08), leaning toward harm but not significant.
• FORWARD, 586 people. 1 gram a day for a year. Recurrence was 24.0 versus 18.9 percent (HR 1.28, 95% CI 0.90 to 1.83), no significant difference.
• AFFORD, 337 people. 4 grams a day of fish oil. Recurrence was 64.1 versus 63.2 percent (HR 1.10), no significant difference.
• A cardioversion trial of 204 people. 3 grams a day before cardioversion and 2 grams after. AF returned in 58.9 versus 51.1 percent (P=0.28), no significant difference. Its authors included the lead authors of the positive 2005 surgery trial.

One smaller placebo-controlled trial did report a benefit. A 2011 Italian trial enrolled 199 people with persistent AF, all of whom also took amiodarone, a heart rhythm drug, and a blood pressure drug of the ACE inhibitor or ARB type. According to the trial's registry results, 61 of 100 people on 2 grams a day of Omacor, a prescription omega-3, stayed in normal rhythm for a year after cardioversion, compared with 34 of 99 on an olive oil placebo. Other trials have not reproduced it: the 204-person cardioversion trial above, in which antiarrhythmic drugs were chosen by local doctors, and AFFORD, in people not taking antiarrhythmic drugs, found no significant benefit.

In 2017, before the large safety trials reported, the American Heart Association concluded that omega-3 supplements offer no benefit for preventing recurrent AF or AF after heart surgery, and made no recommendation on preventing a first episode because no large placebo-controlled trials existed. The prescription label for Lovaza, an omega-3 ethyl ester drug, now notes a possible association with more frequent recurrences in people with paroxysmal or persistent AF, particularly in the first months of treatment. The figures printed on that label for its underlying trial differ from the published paper's, so we quote the published numbers above.

Fish on the plate versus capsules

Studies that follow people over time, rather than assigning them a treatment, point in a different direction, at least for omega-3 from food.

Fish and blood levels. In the US Cardiovascular Health Study of 4,815 adults aged 65 and older, those who ate tuna or other broiled or baked fish one to four times a week had 28 percent less AF over 12 years than those who ate it less than once a month (HR 0.72), while fried fish showed no association. A 2023 pooled analysis of 17 cohorts, 54,799 people with 7,720 cases of AF, measured omega-3 in blood or tissue: higher DHA (HR 0.90, comparing the 90th with the 10th percentile) and higher EPA plus DHA (HR 0.93) went with less AF, while EPA alone showed no significant association (HR 1.00). One author has an ownership interest in an omega-3 supplement company and another holds stock in an omega-3 blood testing laboratory. Not every study agrees. In a Danish cohort, people in the top fifth of fish omega-3 intake had more AF (adjusted HR 1.34).

Supplement users in the UK Biobank. Three analyses of the same half-million-person UK cohort illustrate how fragile observational findings can be. A 2022 analysis found that people who reported regularly taking fish oil had slightly more AF (6.2 versus 5.2 percent, adjusted HR 1.10), but only among those without cardiovascular disease at the start (5.3 versus 4.1 percent); among those with cardiovascular disease there was no significant difference (11.6 versus 11.1 percent). That is the reverse of the trial pattern, where the signal was clearest in high-risk people. A 2024 analysis of people free of cardiovascular disease reported a similar HR of 1.13. A 2025 reanalysis that adjusted for age as a continuous number, rather than splitting people into two age groups, found no significant association (HR 1.00, 95% CI 0.97 to 1.02), and found that higher blood omega-3 levels went with less AF (HR 0.89). Its authors include the chief medical officer of an omega-3 supplement company and a stockholder in an omega-3 testing laboratory. A 2026 analysis of the same cohort, using a different statistical approach, linked higher total omega-3 levels and supplement use to less AF, but higher DHA levels to slightly more new AF, a pattern not evident in women. None of these analyses knows what dose or product people took.

Genetics. A Mendelian randomization study, which uses genetic variants that raise blood fatty acid levels as a natural experiment, drew on 65,446 people with AF and found that genetically higher levels of any of ten fatty acids, including EPA and DHA, were not associated with AF risk.

The likely explanations for the gap between trials and observational studies are dose, since diet rarely supplies grams of EPA a day; the healthy habits of people who eat fish; and the possibility that EPA and DHA behave differently. A 2023 review coauthored by one of the OMEMI investigators concluded that Clinicians should inform patients that marine omega-3 fatty acid supplement may increase AF risk, while noting the opposite association for blood levels.

Why might omega-3 raise AF risk?

Nobody knows. A 2024 review in Circulation states plainly that the mechanism behind the increased AF is unknown, while noting that the signal appeared with both purified EPA and a mixed EPA plus DHA drug. There are clues:

• Changes in the heart's electrical timing. In a VITAL substudy of 911 people who had ECGs at the start and two years later, 840 mg a day lengthened the PR interval and the P wave, the parts of the tracing that reflect electrical travel through the upper chambers, and raised a measure of heart rate variability linked to vagal nerve activity. The authors suggested these could reflect higher vagal tone or slower conduction within the atria. Researchers have proposed that such changes could make AF easier to trigger in some people, but that is a hypothesis.
• EPA levels. In OMEMI, the rise in blood EPA statistically explained about 65 percent of the treatment's effect on AF risk, while the much smaller rise in DHA did not; in an earlier analysis, DHA rises showed a similar but nonsignificant tendency toward AF (HR 1.39, 95% CI 0.90 to 2.13). Larger rises in EPA went with both more AF (adjusted HR 1.36) and fewer major heart events (HR 0.86), a trade-off in a single measure.
• A lower heart rate. Fish oil slows the resting heart rate slightly, another sign of effects on the heart's pacing system.

The choice of placebo is an unlikely full explanation. AF rose with three different comparisons at 1.8 grams or more: mineral oil in REDUCE-IT, corn oil in STRENGTH and OMEMI, and no placebo at all in RESPECT-EPA. Olive oil appeared only in the 1 gram trials, though, so dose and comparator cannot be fully separated.

What regulators and guidelines say

United States. Since December 2019, the Vascepa label has warned that the drug is associated with an increased risk of AF or atrial flutter requiring hospitalization, citing the REDUCE-IT figures, and says the incidence was greater in people with a previous history of AF. It lists AF among the drug's common side effects. The Lovaza label carries the recurrence caution described above. The FDA has no comparable requirement for fish oil supplements.

Europe and the UK. In September 2023, the European Medicines Agency's safety committee added AF as a common side effect for medicines containing omega-3-acid ethyl esters, which are prescribed for high triglycerides. It described a dose-dependent risk in people with cardiovascular disease or risk factors, highest at 4 grams a day, and advised that treatment be stopped permanently if AF develops; it judged a causal link to be at least a reasonable possibility. The UK's medicines regulator adopted the same wording in January 2024, and added two points worth knowing. It had received no reports of AF linked to omega-3 medicines through its Yellow Card side effect reporting scheme up to 13 November 2023. And it said that neither the trials nor its review looked at fish in the diet or at supplements, so it could not give advice on the risk for people taking omega-3 supplements who do not have known cardiovascular disease or major risk factors.

Japan. In November 2024, Japan's drug regulator revised the labels of icosapent ethyl, sold there as Epadel, and omega-3-acid ethyl esters, sold as Lotriga, to list AF and atrial flutter among their serious adverse reactions, citing REDUCE-IT, STRENGTH and RESPECT-EPA. Because RESPECT-EPA used 1.8 grams a day, Japan's warning rests partly on evidence below 4 grams.

Supplements. All of these warnings apply to prescription medicines. Health Canada's monograph for licensing fish oil as a natural health product requires no statement on adverse reactions and carries no AF warning. The NIH Office of Dietary Supplements fact sheet notes that two large trials of 4 grams a day slightly increased AF risk in people with or at high risk of cardiovascular disease, although it calls those products supplements when both were prescription drugs.

Cardiology guidelines. The 2023 US atrial fibrillation guideline notes that blood omega-3 levels are inversely related to AF, but that with supplementation there has been no effect or a potential for greater AF, and it makes no recommendation for omega-3 in AF prevention. The 2023 US guideline for chronic coronary disease notes that incident AF was more common with both icosapent ethyl and the carboxylic acid formulation. It recommends against nonprescription supplements, including fish oil, for people with chronic coronary disease because they have not been shown to reduce cardiovascular events, and states that omega-3 supplements are not acceptable substitutes for icosapent ethyl. The 2024 European AF guideline, as far as we can find, does not address fish oil.

What this means for your dose

The single most useful thing you can do is read the back of the label. The front usually shows the total fish oil per capsule; the number that matters is the EPA plus DHA. According to the NIH Office of Dietary Supplements, a typical 1,000 mg fish oil capsule provides about 180 mg EPA and 120 mg DHA, or about 300 mg in total. By that yardstick:

• VITAL and ASCEND, 840 mg of EPA plus DHA: close to three typical capsules. No significant difference in AF.
• OMEMI, about 1.6 grams of EPA plus DHA: about five typical capsules. More AF, not quite statistically significant.
• RESPECT-EPA, 1.8 grams of prescription EPA: the EPA in about ten typical capsules. More AF.
• Lovaza at 4 grams: about 3.4 grams of EPA plus DHA ethyl esters according to its label, the amount in about eleven typical capsules.
• Vascepa at 4 grams: 4 grams of purified EPA ethyl ester, roughly the EPA in twenty typical capsules. More AF.

Concentrated products change the math. Some high-potency fish oils deliver more than 1,000 mg of EPA plus DHA per serving, so two servings a day can approach the range where the trial signal becomes clearer, especially for someone with heart disease. The FDA has concluded that supplements providing no more than 5 grams a day of EPA and DHA are safe when used as recommended; the AF data show why, for people with heart disease, a dose well below that ceiling is still worth discussing with a cardiologist. This is a description of trial doses and labels, not a dosing recommendation. Our fish oil buying guide shows how to read EPA and DHA per serving, and our omega-3 fatty acids page covers the other evidence.

Who should be careful

People who already have AF. In REDUCE-IT, people with a history of AF were hospitalized for it about twice as often on the drug as on placebo, and the Lovaza label cautions about more frequent recurrences. European regulators tell prescribers to stop omega-3-acid ethyl ester medicines permanently if AF develops, while the UK regulator adds that the decision needs clinical judgment and that patients should not stop treatment without first talking to their doctor; the European label for icosapent ethyl asks for monitoring instead. For fish oil more generally, placebo-controlled trials in people who already had AF, including FORWARD at 1 gram a day and AFFORD at 4 grams, found no significant difference in recurrence, although recurrence was numerically higher on fish oil in both. No large trial has shown a benefit, which is a reason to involve your cardiologist rather than a reason for reassurance.

People with heart disease, heart failure or high cardiovascular risk. Every statistically significant AF result in the trials came from these groups, and the 2026 meta-analysis found a statistically significant increase only in high-risk people taking high doses. Older age was among the predictors of AF on omega-3 in the STRENGTH analysis, although the same six-factor risk score predicted AF about as well in the placebo group, and OMEMI enrolled people aged 70 to 82.

People taking blood thinners. Most people with AF take an anticoagulant to prevent stroke, so bleeding is the other half of the question. The evidence is mixed. The Vascepa label states that the drug is associated with an increased risk of bleeding (any bleeding 12 versus 10 percent in REDUCE-IT), while serious bleeding in the trial report was 2.7 versus 2.1 percent (P=0.06), not a significant difference. The extra bleeding concentrated in people also taking blood thinners: according to the European label for icosapent ethyl, serious bleeding in REDUCE-IT was more common on the drug among people taking antithrombotic medicines such as aspirin, clopidogrel or warfarin (3.4 versus 2.6 percent) and occurred in 0.2 percent of both groups among those who were not. Among REDUCE-IT participants who already had AF, 77.8 percent of whom took an oral anticoagulant, serious bleeding was 7.3 versus 6.0 percent, no significant difference. A 2024 meta-analysis of 11 trials and 120,643 people found no significant difference in bleeding overall with omega-3 (rate ratio 1.09, 95% CI 0.91 to 1.31), but high-dose purified EPA raised the relative risk of bleeding by about 50 percent, an absolute increase of 0.6 percentage points. For warfarin, a small placebo-controlled study of 3 or 6 grams a day found no significant difference in INR, the clotting test used to dose warfarin, and a clinic study of 573 warfarin users found no significant difference in INR control or bleeding among those taking fish or krill oil. But one case report describes INR rising from 2.8 to 4.3 after a patient doubled her fish oil from 1 to 2 grams a day. We found no study of fish oil combined with the newer anticoagulants such as apixaban or rivaroxaban. The prescription omega-3 labels call for periodic INR checks when these products are taken with anticoagulants. Our guide to supplement and drug interactions covers the questions to ask.

Know the symptoms. The Vascepa patient information lists symptoms of AF and atrial flutter to watch for: a fast or irregular heartbeat, lightheadedness, dizziness, shortness of breath, chest discomfort and fainting. These deserve prompt medical attention whatever you take.

Frequently asked questions

Does fish oil cause atrial fibrillation?

At high doses in people with heart disease or high cardiovascular risk, randomized trials found more atrial fibrillation in people taking omega-3 than in those taking a placebo or no added treatment. The clearest results came from prescription omega-3 drugs at 4 grams a day, and a Japanese trial of 1.8 grams a day of prescription EPA. The absolute increase was small for most participants, about 1 percentage point over several years, but larger in people who already had AF (12.5 versus 6.3 percent hospitalized for AF in REDUCE-IT). At about 1 gram a day, the largest trial, VITAL Rhythm, found no significant difference from placebo, although pooled analyses cannot rule out a small increase at that dose.

How much fish oil is too much for heart rhythm?

No trial has identified a safe threshold. In one 2021 meta-analysis, risk rose by about 11 percent for each extra gram a day of omega-3 product (counted by capsule weight, not EPA plus DHA content), and a 2026 meta-analysis found a statistically significant increase only in high-risk people taking more than 1,500 mg a day of EPA plus DHA. Most single-capsule products provide far less than that, but concentrated products and multi-capsule servings can reach it. Read the EPA plus DHA amount on the back label, and if you have heart disease or atrial fibrillation, discuss any dose above what a single standard capsule provides with your cardiologist.

Should people with atrial fibrillation stop taking fish oil?

That is a decision to make with your cardiologist, not on your own, and it should never involve changing a prescribed medicine. The evidence is relevant, though. In REDUCE-IT, people who already had atrial fibrillation were hospitalized for it about twice as often on the high-dose drug as on placebo, and the Lovaza prescription label notes a possible link to more frequent recurrences in people with paroxysmal or persistent atrial fibrillation. No large placebo-controlled trial has shown that fish oil reduces atrial fibrillation in people who have it; two small trials, one of them without a placebo, reported fewer recurrences, but larger placebo-controlled trials did not confirm them.

Is eating fish safe for people worried about atrial fibrillation?

The signal comes from supplements and prescription drugs at doses far above what diet usually provides. In observational studies, eating broiled or baked fish, and higher blood levels of omega-3 (which reflect both diet and supplements), were linked to the same or lower rates of atrial fibrillation, not higher rates. One Danish cohort found more atrial fibrillation at the highest intake, so the picture is not uniform, but fish as part of a normal diet has not shown the pattern seen with high-dose capsules.

Is EPA or DHA responsible for the atrial fibrillation risk?

It is not settled. The increase appeared both with purified EPA (REDUCE-IT and RESPECT-EPA) and with an EPA plus DHA drug (STRENGTH). In the OMEMI trial, the rise in blood EPA statistically explained most of the extra atrial fibrillation while the much smaller rise in DHA did not, and a pooled analysis of 17 cohorts linked higher DHA levels to lower risk, although a 2026 UK Biobank analysis found a modest link between higher DHA and new atrial fibrillation that was not evident in women. That hints EPA matters more, but no trial has compared the two directly for atrial fibrillation.

Can fish oil prevent atrial fibrillation?

The larger trials do not support it. Small early studies, mostly without a placebo, reported less atrial fibrillation after heart surgery and after cardioversion, but larger placebo-controlled trials found no significant benefit, and several leaned toward more atrial fibrillation. The American Heart Association concluded in 2017 that omega-3 supplements offer no benefit for preventing recurrent atrial fibrillation or atrial fibrillation after heart surgery.

Can I take fish oil with a blood thinner?

Only with your prescriber's knowledge. Most people with atrial fibrillation take an anticoagulant such as warfarin or apixaban. A meta-analysis of 11 trials found no significant difference in bleeding overall with omega-3, but high-dose purified EPA raised bleeding risk by about 0.6 percentage points, and the Vascepa label reports more bleeding in people also taking antithrombotic drugs. Two warfarin studies, a placebo-controlled trial that enrolled 16 patients and a retrospective clinic study of 573 users, found no significant effect on INR or INR control, but one case report describes a rise in INR after a patient doubled her fish oil dose, and we found no study of fish oil with the newer anticoagulants. The NIH Office of Dietary Supplements notes that the prescription omega-3 labels call for periodic INR monitoring when these products are taken with anticoagulants.

Is krill oil or algae oil safer for heart rhythm?

No trial has been designed to test krill oil or algae oil for effects on atrial fibrillation, so there is little evidence either way. The pattern across the fish oil trials appears to follow dose and cardiovascular risk more than the specific product, so the sensible assumption is that the total EPA plus DHA you take matters most, whatever the source.

The bottom line

The heart trials changed the story of fish oil and heart rhythm. At high doses, mostly prescription omega-3 drugs at 1.8 to 4 grams a day in people with heart disease or high cardiovascular risk, AF became more common, a finding now written into drug labels in the US, Europe, the UK and Japan. The increase is real but small in absolute terms for most participants, around 1 percentage point over several years, though larger in people who already had AF. REDUCE-IT also reported fewer strokes on its prescription drug, but STRENGTH found no significant difference in nonfatal stroke, and neither result shows that a supplement lowers stroke risk. At about 1 gram a day, the largest and most carefully measured trial found no significant difference, though the pooled data cannot rule out a small effect. In the larger placebo-controlled trials that tested fish oil for preventing AF, there was no significant benefit, and fish in the diet has not shown the increase seen with high-dose capsules. If you are healthy and take a standard capsule, the evidence gives little reason for alarm. If you have AF or heart disease, take a blood thinner, or are considering a high-dose product, treat omega-3 as something to decide with your cardiologist, and count every capsule when you do.