Fish Oil Benefits: Little or No Help for Most Hearts. The Exceptions Are Narrow

There is a genuine, well-evidenced benefit to omega-3. The American Heart Association wrote a science advisory recommending it.

It requires four grams a day, a prescription, and a blood test result most people buying fish oil have never had. [2]

The capsule in your cupboard is a different proposition, and it has been tested about as thoroughly as anything in nutrition.

One fence before the evidence: this page rates the heart claim. Omega-3 in pregnancy is a separate question with a different answer — a 2018 Cochrane review found about 11% fewer births before 37 weeks, across 26 trials, and rated that high-quality evidence [6] — and we have not rated it here.

What 86 trials and 162,796 people found

In 2020 Cochrane — the organisation that exists to pool trials without a commercial stake in the answer — updated its review of omega-3 and cardiovascular disease. Eighty-six randomised trials. Doses from half a gram to more than five grams a day. Follow-up from one year to more than seven. [1]

Look at the right-hand column, which none of the nine first-page web results we read for “fish oil benefits” on 7 October 2026 gives. Certainty is a separate judgement from the result: it is how much confidence the evidence itself deserves, given how the trials were run and how consistent they were.

On dying of anything, and on cardiovascular events, the answer is little or no difference — at high certainty. That is the strongest grade Cochrane issues. It does not mean the question is closed forever, but it does mean more trials of the same kind are unlikely to change the answer.

Two small signals survived, both at low certainty. Coronary heart disease deaths came out about 10% lower, with a number needed to treat of 334; coronary heart disease events about 9% lower, with a number needed to treat of 167. That last figure means: if 167 people take fish oil, one of them avoids an event who otherwise would not have. The other 166 get nothing.

Cardiovascular death as a whole came out 8% lower on paper. The reviewers read that same evidence as “probably makes little or no difference”: in the best-run trials the effect moved closer to none at all, and the grade stops at moderate rather than high only because the range of plausible answers still takes in a benefit worth having. [1]

And one line from the review deserves its own paragraph, because it undercuts the advice everybody gives: “There is little evidence of effects of eating fish.” [1]

The biggest trial of the capsule people actually buy

VITAL is the one designed to answer the consumer question directly: 25,871 healthy older adults, a one-gram fish oil capsule a day carrying 840 mg of EPA and DHA, the two omega-3s fish oil is sold for, versus placebo, for a median of 5.3 years. [3]

Major cardiovascular events: 386 on omega-3, 419 on placebo. Hazard ratio 0.92, and a p-value of 0.24 — comfortably inside the range you get from chance. Cancer: no difference either.

There is one result in VITAL worth being careful with. Total heart attacks came out significantly lower on omega-3, a hazard ratio of 0.72. [3] That is a real number and an interesting one.

It is also a secondary endpoint inside a trial whose primary endpoint was null. Trials measure many things; some of them come out significant by chance, which is precisely why one outcome is nominated in advance as the one that counts. When the headline result is negative and a sub-result is positive, the honest description is “worth following up”, not “fish oil prevents heart attacks”.

So what is the real benefit?

Triglycerides. A type of fat in your blood, separate from cholesterol, and high levels are their own cardiovascular risk factor.

The American Heart Association’s 2019 advisory is unambiguous: at 4 grams a day, omega-3 agents reduce triglycerides by 30% or more in people with very high levels. In the milder range, EPA-only and EPA-plus-DHA work about equally and neither raises LDL cholesterol. [2]

Note the dose. The advisory is explicitly about pharmacological amounts — above 3 grams a day of combined EPA and DHA. A standard supermarket fish oil capsule contains a fraction of that, and the label figure is usually total fish oil rather than actual EPA and DHA, which makes the gap wider than it looks.

There is also a trial showing hard outcomes at that dose. REDUCE-IT randomised 8,179 patients to 4 grams a day of a purified prescription EPA drug or placebo, and cut major cardiovascular events from 22.0% to 17.2% — a 25% relative reduction. [4]

Every word of the entry criteria matters, though. Those patients had established cardiovascular disease or diabetes with risk factors, were already taking a statin, and were selected for elevated triglycerides. It is a prescription drug trial in high-risk patients. It is not evidence for a gram of fish oil in a healthy forty-year-old, and it is regularly cited as though it were.

Its result has not been repeated with the other kind of prescription omega-3. STRENGTH gave 13,078 statin-treated, high-risk patients with high triglycerides 4 grams a day of an EPA-plus-DHA prescription, against corn oil. It was stopped early, when an interim look found little chance of benefit: 12.0% had a major cardiovascular event on the drug, 12.2% on corn oil. [7]

REDUCE-IT’s placebo is the other open question. It was mineral oil, chosen to match the drug’s colour and consistency, and the trial’s own paper argued that the small cholesterol difference between the groups was unlikely to explain a 25% benefit. [4] A later substudy, paid for by Amarin, which makes the drug, reported that in the first year, among people taking the mineral oil, LDL cholesterol rose about 11% and C-reactive protein, a blood marker of inflammation, about 22%, while the drug group barely moved. Its authors called the effect of that on the headline result “uncertain”. [8]

Both sides have since put numbers on it. A Danish team took the cholesterol, triglyceride and inflammation changes seen in both arms of REDUCE-IT and asked what they should have done to heart risk, using 5,684 people from a Copenhagen population study who would have qualified for the trial. Those changes predicted about a 12% lower risk, where the trial found 25%, and the team concluded the different placebos explain part of the gap between REDUCE-IT and STRENGTH, but not all of it. [9] A review of 80 studies that used mineral oil as a placebo found its effects on those markers inconsistent and concluded it does not meaningfully change the result; six of its ten authors were Amarin employees. [10]

Nobody has settled it, and this page will not pretend to. What it does mean is that the positive one of the two 4-gram prescription trials has an open question over its comparison group, and the trial with a different drug and a different placebo found nothing. Both trials were paid for by the companies whose drugs they tested; one found a benefit and one did not.

Fish oil for people on dialysis: a large trial found fewer heart events

Has fish oil cut heart events in anyone, then? In a few large trials it has, and each comes with a catch. GISSI-Prevenzione, the 1999 trial at the start of this page’s provenance chain, had no placebo [26]. REDUCE-IT tested a prescription EPA drug at 4 grams, in people picked for high triglycerides [4]. JELIS, a Japanese trial of an EPA drug at 1.8 grams, which the verdict below describes, had no placebo [21]. And GISSI-HF gave 6,975 people with heart failure in Italy a one-gram capsule a day, carrying 850 to 882 mg of EPA and DHA, or a placebo, for a median of 3.9 years: 27% died on the capsule against 29% on placebo, which its authors called “a small beneficial advantage”. Its catch is size: by its authors’ reckoning, 56 people had to take the capsule for nearly four years to avoid one death [22]. All four are among the 86 trials in the Cochrane review above, which still found little or no difference to deaths or to cardiovascular events overall [1].

A newer trial is a different case: a high-dose fish oil concentrate against a placebo, in people on hemodialysis, the machine treatment that does the kidneys’ job of cleaning the blood once they have failed.

The trial is called PISCES, and the New England Journal of Medicine published it in November 2025. It took adults on hemodialysis three or four times a week at 26 dialysis units in Canada and Australia. Every day they swallowed four 1-gram capsules of a fish oil concentrate, which the trial calls 4 grams of omega-3: 1.6 grams of EPA and 0.8 of DHA, 2.4 grams of the two in all. That is nearly three times the 840 mg in VITAL’s daily capsule [3] (our arithmetic). It was not a prescription drug. The comparison was matching capsules of corn oil, and neither the patients nor the trial staff knew who had which [13] [14].

Who were they? Their average age was 64, about six in ten were men, and about two-thirds had no history of heart or blood-vessel disease. A little over half took a statin. Their triglycerides were unremarkable, with a middle value of about 114 to 120 mg/dL (1.29 and 1.35 mmol/L) in the two groups, where REDUCE-IT and STRENGTH had picked people with high levels. People already taking fish oil, people with an implanted defibrillator, and people taking more than two blood-thinning or antiplatelet drugs were left out [13] [14].

The trial planned to follow them for three and a half years; a quarter of them were followed for a year or less. Serious cardiovascular events (cardiac death, heart attack, stroke, and circulation problems in the legs bad enough to need an amputation) came at about half the rate on fish oil: 158 events against 309, or 0.31 against 0.61 per 1,000 patient-days, a patient-day being one person followed for one day. The hazard ratio, the trial’s comparison of risk where 1 means no difference, was 0.57, about 43% fewer, with a range of plausible answers from 0.47 to 0.70. Every part of that count moved the same way: cardiac deaths 63 against 113, heart attacks 49 against 96, amputations 35 against 66, strokes 11 against 34 [13].

So is fish oil a heart drug for people on dialysis? Three things stop us short of saying so.

First, the headline counts every event, not just each person’s first: someone who had a heart attack and later a stroke counts twice. That is the count the protocol named [14], and the trial’s registry entry shows the main measure was narrowed to it in August 2020, while the trial was running, from counting first events as well. Counted the other way, by each person’s first event, the answer came out much the same, a hazard ratio of 0.56 [13].

Second, deaths from any cause were not significantly fewer: 175 on fish oil against 195 on corn oil, a hazard ratio of 0.89, with a range of plausible answers the paper prints as 0.73 to 1.01. That upper end may be a misprint, on our reading. The paper says deaths were compared with a standard survival model, which gives a range balanced in proportion either side of its estimate: with 0.89 in the middle and 0.73 at the bottom, the top should sit near 1.09, which is also what the death counts give with the usual approximation (our arithmetic). The trial’s supplementary appendix, using a different model, gives deaths 0.90 with a range up to 1.13 [13]. Either way, the range takes in no difference at all. Cardiac deaths fell, 63 against 113. Other deaths, counted raw, went the other way: the non-cardiac deaths that the main measure did not already count came to 108 on fish oil against 72 on corn oil (our subtraction from the trial’s table: 266 minus 158, and 381 minus 309) [13]. Two of the trial’s secondary measures count deaths from every cause alongside heart events, and both favoured fish oil: the main measure with those other deaths added, a hazard ratio of 0.77 (0.65 to 0.90), and a first heart event or death from any cause, 0.73 (0.61 to 0.87) [13]. The paper adds that the ranges for all its secondary measures, these two and deaths among them, were not widened to allow for the extra chance of a fluke that comes with testing many outcomes, and should not be used to decide whether a difference is real [13].

Third, it rests on one large trial. The newest pooling we found, published in June 2026, describes its 13 trials as tests of fish oil or omega-3 against a placebo in people on hemodialysis; PISCES holds 1,228 of their 2,755 people and 85% of the weight in the pooling’s estimate for cardiovascular deaths [16]. Two of the 13 fit that description loosely. The next largest, 567 people, gave fish oil or an olive-oil placebo for 12 weeks around the operation that makes the blood-vessel connection used for dialysis, to people with advanced kidney disease, about half of them not yet on dialysis, and tested whether the connection kept working [23]. And one, below, had no placebo.

Three smaller trials in the pooling had set out to count heart events, going by their titles, and they did not all point the same way. One, by the same team as PISCES, gave 201 people with new access grafts a similar capsule or corn oil for a year; they stayed free of heart events longer on fish oil, a secondary result with a wide range of plausible answers, a hazard ratio of 0.43 (0.19 to 0.96) [17]. One, in Denmark, gave 206 people with existing heart or blood-vessel disease 1.7 grams a day of an omega-3 medicine or olive oil for two years, and found no significant effect on its main measure of heart events and deaths, though fewer heart attacks, 4 against 13 [18]. And one, in Japan, published only as a meeting abstract, gave 179 people 1.8 grams of EPA a day or no EPA for two years, open-label, meaning patients and doctors knew who was taking it: heart and blood-vessel events came to 13 against 27, a difference its authors report as significant [24]. The 2026 pooling’s own authors write that its evidence is concentrated in a single large trial, and that this calls for caution [16]. PISCES has not been repeated [19].

What about the side effects this page warns of? Bleeding is the main safety worry with omega-3, the trial’s protocol says [14], and serious bleeding was less common on fish oil: 29 people (4.8%) against 47 (7.6%). Serious infections went the other way, 124 people (20.3%) against 110 (17.8%); the trial reports that adverse events overall did not differ meaningfully [13]. The trial counted bleeding through its reports of serious adverse events, and neither its paper nor its supplementary appendix gives a figure for minor bleeding [13]; a 2026 commentary in a kidney journal says minor bleeding was slightly more common on fish oil, which we could not check against the trial’s paper or appendix [20]. Nor does either report atrial fibrillation, the irregular heartbeat in the next section, as something that happened during the trial: the appendix lists it only among the conditions people had when they joined, 9.0% of the fish-oil group and 8.1% of the placebo group [13].

Who paid? A Canadian heart charity, Australia’s national medical research council, two Canadian hospital research funds, a kidney-care network and a private donation. The capsules and the corn-oil placebo were donated by a company that makes omega-3 ingredients. The trial was run from a Toronto hospital research institute, which its registry lists as a non-commercial sponsor; its protocol declares no competing interests and says the funders had no role in its design, analysis or write-up; and its named authors work at universities and hospitals [13] [14]. In their disclosure forms, seven of the nine authors declare no interests; the other two list ties to drug companies, one of them the maker of the omega-3 drug STRENGTH tested, and neither lists the company that donated the capsules [13].

What do other doctors dispute? The journal ran an editorial beside the trial, by two doctors at hospitals in Boston and New York, under the title “Red Herring or Great Catch?” [25]. Without a subscription we could read only its first paragraph, which sets out how many people are on hemodialysis and how high their death rate is; a 2026 commentary reports that the editorial called the findings “remarkable but unexpected” and urged confirmation before practice changes [20]. The journal also printed three letters and the authors’ reply in July 2026, by subscription. We could read only the opening of the first, from three doctors in Japan, which points out that people’s background intake of the omega-3s in fish differs markedly between regions; the rest of their argument is behind the paywall [15]. The other two letters and the authors’ reply we have not read.

So whom does it apply to? Adults on hemodialysis, at that dose. Not people on peritoneal dialysis or with a working kidney transplant, who could not join, and not the healthy adults of VITAL or the statin-treated patients of REDUCE-IT and STRENGTH [14]. Blood levels of EPA and DHA are often low in people on dialysis, as the trial’s protocol notes [14], and on our reading that is one more reason not to carry the result over to anyone else.

In words: that fish oil cuts serious heart events in people on hemodialysis is Preliminary. The trial is large, carefully blinded and pointed the same way on every part of its main measure. But it is early: one large trial, not yet repeated, that carries most of the weight in the newest pooling; deaths overall were not significantly fewer; and the three earlier trials in the pooling that set out to count heart events were small, one of them without a placebo, and did not all point the same way [13] [16] [17] [18] [24]. This page’s own rating is about fish oil taken for the heart by everyone else, and it does not change.

This is journalism, not medical advice. If you are on dialysis, take a blood thinner, or have been prescribed omega-3, do not start, stop or change a dose without your kidney team or the clinician who prescribed it.

The side effect now on the prescription label

This is the part of the fish oil story that has genuinely changed, and the marketing has not caught up.

In 2021, researchers pooled seven cardiovascular outcome trials covering 81,210 patients and looked specifically at atrial fibrillation — an irregular heart rhythm that raises stroke risk and often needs lifelong treatment. [5]

Omega-3 supplementation was associated with a 25% higher risk of atrial fibrillation. Hazard ratio 1.25, confidence interval 1.07 to 1.46.

And it was dose-dependent, which is the detail that makes it credible rather than a fluke. Above one gram a day: hazard ratio 1.49. At one gram a day or below: 1.12 — smaller, but with a confidence interval of 1.03 to 1.22, which still does not include “no effect”. The test for whether the two doses differed came out at p < 0.001, and a meta-regression found the risk climbing with every extra gram. [5]

We should report the tension here rather than tidy it away. Cochrane’s review found little or no difference in “arrhythmia” as a category, one that includes atrial fibrillation — risk ratio 0.99, though at low certainty. [1] The 2021 analysis asked a narrower question, about atrial fibrillation specifically, in a smaller set of trials: five of its seven are also in Cochrane’s pool, and the other two reported after Cochrane’s search. Both are on the record. The narrower, dose-responsive one is the one we would want a doctor to know about.

A larger analysis has since gone further. Published in 2026, it pooled 35 trials and 114,592 people, including atrial fibrillation counts that some trials had never published. It found a clear increase in one group only: people already at high cardiovascular risk taking more than 1.5 grams a day, where about 8 more in every 1,000 developed atrial fibrillation. At lower doses, and in lower-risk people, it found no increase it could tell apart from chance. The project was part-funded by the omega-3 industry’s own advocacy group, and three of its authors disclosed ties to omega-3 businesses. [11]

Put the two atrial fibrillation analyses side by side and they agree at the top of the dose range and disagree at the bottom. The top is where regulators have acted. In September 2023 the European Medicines Agency’s safety committee added atrial fibrillation as a common side effect of omega-3-acid ethyl ester medicines, the prescription capsules for high triglycerides, and said the risk is highest at 4 grams a day. [12]

The verdict

Fish oil supplements do not protect your heart, and that finding carries Cochrane’s highest certainty grade on both dying and cardiovascular events. The largest trial in healthy adults missed its primary endpoint. Small signals on coronary events survive at low certainty and a number needed to treat of 167.

Against that, more atrial fibrillation, and the risk rises with the dose. Above a gram a day, the 2021 pooling found 49% more. At a gram a day or less it found 12% more, and the larger 2026 analysis found no increase below 1.5 grams that it could tell apart from chance.

The real benefit is narrow and it is genuine: at four grams a day, omega-3 lowers triglycerides substantially, and the American Heart Association recommends it for exactly that. If you have high triglycerides, that is a conversation with your doctor about a prescription — not a reason to buy a bigger tub. Ask which one, too: of the two 4-gram prescription trials, the one that cut heart events used EPA alone and the EPA-plus-DHA one found nothing, and the atrial fibrillation risk is highest at the 4-gram dose. A third trial also tested EPA alone, at a lower dose, with its own catch. JELIS gave 18,645 Japanese patients with high cholesterol, a group it describes as eating a lot of fish, 1.8 grams a day of an EPA drug on top of a statin, or the statin alone, with no placebo, though its events were judged blind. It found 19% fewer major coronary events (2.8% against 3.5%), “especially” non-fatal ones; sudden cardiac death and coronary death “did not differ between groups”, the paper says [21]. Its registry names the drug’s maker as a collaborator, and the Cochrane review records that company as its funder; the paper’s own funding statement is behind a subscription and we have not read it [1] [21]. But the two Italian trials above gave EPA with DHA in a one-gram capsule and also reported fewer deaths or heart events, each with its own catch, and so did the dialysis trial, at a higher dose. Together, these trials do not settle whether EPA alone does better than EPA with DHA [26] [22] [13].

If you do not have high triglycerides, the honest summary is that you are taking a supplement with a high-certainty null on the outcome you care about and a dose-dependent signal, faint or absent at supplement doses, on one you have probably never considered.

People on hemodialysis are a separate case, resting mostly on one large trial. There, a high-dose fish oil concentrate, not a prescription drug, cut serious heart events by about 43% against corn oil, without significantly fewer deaths overall, and we rate that Preliminary until it is repeated. If that is you, take the question to your kidney team rather than to the supplement aisle [13].

Choosing between fish oil and krill oil? We checked that one too. The review we read pooled 64 trials, and only four of them put the two oils in the same trial, each with a krill-oil company involved; pooled with the wider evidence, no difference in triglycerides or cholesterol held up.

Related: fish oil vs krill oil · do testosterone boosters work · ashwagandha benefits · berberine benefits · how to increase testosterone naturally