The short version: the number on the front of the bottle is not the dose. A standard "Fish Oil 1000 mg" capsule contains about 300 mg of EPA+DHA — so the 4 g/day dose used in the only positive cardiovascular outcome trial is thirteen capsules a day. Nobody takes thirteen capsules a day.

Meanwhile three regulators — the EMA, the UK MHRA and the US FDA — have added atrial fibrillation to the label of prescription omega-3. We could find no equivalent warning on any over-the-counter Australian fish oil, and the Heart Foundation's guidance document on fish and omega-3 is dated 2015 — before every major trial discussed below.

And one tin of sardines beats a day's capsules on dose and on price.


What fish oil is genuinely good for

ClaimEvidenceVerdict
Lowers triglyceridesCochrane, high certainty; AHA advisoryReal, dose-dependent — needs 4 g/day for 20–30%
Prevents heart attack, stroke or death in the general populationCochrane, high certainty, nullNot supported
Reduces events in high-risk statin-treated patients with high triglyceridesREDUCE-IT only; placebo disputed; STRENGTH nullContested, prescription-only
Reduces NSAID use in rheumatoid arthritisOne significant outcome of five, n=370, ≥2.7 g/dayWeak but plausible
Treats depressionCochrane very low certainty, "unlikely clinically meaningful"; VITAL-DEP nullMarketing
Prevents preterm birthCochrane positive but pre-ORIP; ORIP null overallOverturned; possible subgroup, unproven
Infant brain and eye developmentCochrane: "very few differences"Marketing
Dry eyeDREAM nullMarketing
Cognition and dementiaCochrane plus AREDS2, nullMarketing

Triglycerides — the one strong effect

The AHA Science Advisory (Skulas-Ray AC, Wilson PWF, Harris WS, et al. Circulation 2019;140(12):e673–e691) puts the prescription-dose effect at:

  • Moderate hypertriglyceridaemia (2.26–5.64 mmol/L): triglycerides down 20–30%, without raising LDL-C
  • Severe (≥5.65 mmol/L): down more than 30%

Both figures require 4 g/day of EPA+DHA. Cochrane's whole-literature estimate at typical supplement doses is a ~15% reduction, dose-dependent, high certainty — real, but a long way from the advisory's headline.

Cardiovascular outcomes — the high-certainty null

**Abdelhamid AS, Brainard J, Deane KHO, et al. Cochrane Database Syst Rev 2020;(3):CD003177 (pub5). 86 RCTs, n=162,796**, 28 at low risk of bias, evidence to February 2019.

  • All-cause mortality: little or no effect — HIGH certainty
  • Cardiovascular events: little or no effect — HIGH certainty
  • Stroke: no effect — MODERATE certainty
  • Coronary death: number needed to treat ≈ 334; coronary events: NNT ≈ 167 — both LOW certainty
  • Triglycerides down ~15% — HIGH certainty

Cochrane's own press release was headlined that the evidence "challenges belief that omega 3 supplements reduce risk of heart disease, stroke or death." The omega-3 industry has published rebuttals arguing the review is flawed; that is worth knowing, and it does not change the GRADE rating.

Plant omega-3 (ALA) fared slightly better on softer endpoints — CV events NNT ≈ 500, arrhythmia NNT ≈ 91 — at low certainty.

So: for an Australian without very high triglycerides and without established atherosclerotic disease, high-certainty evidence says over-the-counter fish oil does not reduce death, heart attack or stroke.


The one positive trial, and why it is disputed

REDUCE-IT

**Bhatt DL, Steg PG, Miller M, et al. N Engl J Med 2019;380(1):11–22. n=8,179, icosapent ethyl (purified EPA ethyl ester) 4 g/day, median follow-up 4.9 years, placebo = mineral oil**.

  • Primary composite endpoint: HR 0.75 (95% CI 0.68–0.83), p<0.001
  • CV death: HR 0.80 (0.66–0.98)
  • Triglycerides at one year: −18.3% vs +2.2% placebo
  • AF or flutter: 5.3% vs 3.9%; hospitalisation for AF 3.1% vs 2.1%, p=0.004
  • Serious bleeding: 2.7% vs 2.1%, p=0.06

That is a 25% relative risk reduction — the best result in the field. Two things qualify it.

STRENGTH, which found nothing

**Nicholls SJ, Lincoff AM, Garcia M, et al. JAMA 2020;324(22):2268–2280. n=13,078, omega-3 carboxylic acid (EPA+DHA) 4 g/day, comparator corn oil**, median ~41 months.

  • Primary endpoint: 12.0% vs 12.2%, HR 0.99 (0.90–1.09), p=0.84
  • Stopped early for futility
  • AF: 2.2% vs 1.3%, HR 1.69 (1.29–2.21), p<0.001
  • Triglycerides −19.0%; hsCRP −20.0% — the same biochemical effects as REDUCE-IT, and no clinical benefit
  • GI adverse events 24.7% vs 14.7%; diarrhoea 11.9% vs 4.9%

Same dose. Same triglyceride drop. Opposite result.

The mineral oil argument

A banked-sample biomarker analysis (Ridker PM, et al., Circulation 2022) reported that in REDUCE-IT's mineral oil placebo arm, inflammatory markers rose over 12–24 months:

  • IL-1β +28.9%
  • hsCRP +21.9%
  • Lp-PLA2 +18.5%
  • IL-6 +16.2%
  • Oxidised LDL +10.9%

The icosapent ethyl arm was largely flat. LDL-C also rose in the placebo arm, and triglycerides rose 2.2% — against a background of statin therapy, where drift is not expected.

The argument: if mineral oil interfered with statin absorption or was itself mildly pro-inflammatory, part of REDUCE-IT's benefit is placebo harm, not drug benefit. Dariush Mozaffarian put it as "some of the observed benefit wasn't necessarily from the benefit of the fish oil but the harm of the mineral oil."

The rebuttal (Bhatt and the sponsor): the absolute biomarker changes were small and in several cases below the assay's limit of quantification; the benefit tracks the roughly fourfold rise in serum EPA, a plausible dose-response; mineral oil at ~4 mL/day is far below laxative doses; and the FDA reviewed the mineral-oil question and approved the cardiovascular indication anyway in December 2019.

A third explanation nobody can rule out: the two trials used chemically different actives — pure EPA ethyl ester versus an EPA+DHA carboxylic-acid mix. "EPA alone behaves differently from EPA+DHA" is a competing hypothesis with nothing to do with placebos. And corn oil is not inert either; it supplies linoleic acid.

Status: unresolved. Nobody has run the tiebreaker — a 4 g/day EPA trial against a genuinely inert placebo.


The atrial fibrillation signal

This is the part least likely to be on your bottle.

The pooled analysis

**Gencer B, Djousse L, Al-Delaimy WK, et al. Circulation 2021;144(25):1981–1990. 7 trials, n=81,210** (VITAL, ASCEND, GISSI-HF, Risk & Prevention at 1 g/day; OMEMI at 1.8 g; REDUCE-IT and STRENGTH at 4 g).

  • Pooled HR for incident AF: 1.25 (95% CI 1.07–1.46), p=0.013
  • ≤1 g/day: HR 1.12 (1.03–1.22), p=0.024
  • >1 g/day: HR 1.49 (1.04–2.15), p=0.042
  • p for interaction <0.001
  • Per 1 g/day increment: HR 1.11 (1.06–1.15), p=0.001

Note the low-dose stratum was also statistically significant. This is not purely a high-dose phenomenon in the pooled data.

The individual trials, which do not all agree

TrialnDose/dayAF result
VITAL-Rhythm (Albert CM, JAMA 2021;325(11):1061–1073)25,119840 mgHR 1.09 (0.96–1.24), p=0.19 — null
ASCEND (NEJM 2018;379(16):1540–1550)15,480840 mg7.7% vs 7.6% — flat
OMEMI (Kalstad AA, Circulation 2021;143(6):528–539)1,0141,590 mg7.2% vs 4.0%, HR 1.84 (0.98–3.45), p=0.056
REDUCE-IT8,1794,000 mgHospitalisation 3.1% vs 2.1%, p=0.004
STRENGTH13,0784,000 mgHR 1.69 (1.29–2.21), p<0.001

VITAL-Rhythm is the honest counterweight. It is the largest trial, at the dose most Australians actually take, and it found a 9% point estimate with a confidence interval comfortably crossing 1. If you take 1–3 standard capsules a day, that is the most relevant number for you.

There is also a dietary-versus-supplemental divergence. Biomarker and dietary omega-3 studies find neutral-to-protective associations with incident AF, while supplement RCTs find increased risk (see JACC 2023 and the accompanying editorial, "Omega-3 Fatty Acids and Atrial Fibrillation: The Difference Between Dietary Consumption and Supplementation"). We did not extract those effect sizes and are not quoting them. But the direction matters: eating fish is not the thing that carries this signal.

What regulators have done — and the Australian gap

EMA, Direct Healthcare Professional Communication on omega-3-acid ethyl ester medicines, disseminated 8 November 2023:

  • AF added as a "common" adverse reaction (≥1/100 to <1/10)
  • Dose-dependent, highest risk at 4 g/day
  • "If atrial fibrillation develops, treatment should be permanently discontinued"

UK MHRA, Drug Safety Update, 22 January 2024 (Omacor/Teromeg 1000 mg):

  • AF listed as common, incidence 3.9%
  • Based on three meta-analyses, >80,000 patients
  • The MHRA explicitly states it "does not regulate food or dietary supplements marketed without reference to a medicinal effect/claim" and cannot advise on over-the-counter fish oil

FDA, US Vascepa label: "VASCEPA is associated with an increased risk of atrial fibrillation or atrial flutter requiring hospitalization." 3% vs 2%, HR 1.5, greater in those with prior AF.

EMA, Vazkepa label (same molecule): AF 5.8% vs 4.5%; in those with prior AF history, 12.5% vs 6.3%.

Australia. Vazkepa is TGA-registered (ARTG 380979/380980, sponsor AA-Med Pty Ltd, AusPAR September 2023). Given the EU alignment it is near-certain the Australian Product Information carries the same warning, though we have not read the Australian PI directly and are not quoting it.

What we can say: we could find no TGA safety advisory, label-change requirement or AF warning on over-the-counter fish oil complementary medicines. That is an absence-of-evidence finding, not proof of absence. But the shape of it is clear — the warning attaches to the prescription molecule, not to the supermarket bottle, even though the molecules overlap.

And the Heart Foundation's health-professional Q&A on fish and omega-3 is dated 2015. It predates REDUCE-IT, STRENGTH, VITAL-Rhythm, OMEMI and Gencer entirely, and does not mention AF.

In consumer terms

At ~1 g/day, the best single trial found no significant increase, while the pooled low-dose estimate was HR 1.12 — an absolute risk of roughly 1 per 1,000 per year, in older people already at cardiovascular risk. At 4 g/day it is a different proposition: HR 1.49, and three regulators have put it on a label. If you have palpitations, a history of AF, or you are taking 4 g/day, this is the thing worth knowing.


Dose and label reading — the core skill

A standard "Fish Oil 1000 mg" capsule contains 1,000 mg of oil, of which about 300 mg is omega-3 — typically EPA 180 mg + DHA 120 mg. You are getting 30% of what the front of the bottle shouts.

TrialEPA+DHA/dayStandard 300 mg capsules
Heart Foundation general-population target250–500 mg1–2
VITAL / ASCEND / VITAL-Rhythm / VITAL-DEP840 mg~3
OMEMI1,590 mg~5
Rheumatoid arthritis meta-analysis≥2,700 mg~9
DREAM (dry eye)3,000 mg10
REDUCE-IT / STRENGTH4,000 mg~13

Concentrated products change the arithmetic:

  • Blackmores Omega Triple Super Strength: EPA 540 + DHA 360 = 900 mg per capsule → 4 g/day is ~4.5 capsules, 1 g/day is ~1
  • Ethical Nutrients High Strength Omega-3: EPA 456 + DHA 198 = 654 mg → 4 g/day ≈ 6 capsules. Its own label recommends 1 capsule daily for "cardiovascular health" — 654 mg, below every cardiovascular trial dose including the null ones.

The rule: ignore the number on the front. Find EPA and DHA in the ingredients panel, add them, multiply by capsules per day. If a product states only "omega-3 marine triglycerides 300 mg" without breaking out EPA and DHA — as Nature's Own Odourless does — you cannot compute your dose at all.


Australian prices

Retrieved 1 October 2026 from Chemist Warehouse and Blooms. These are promotional prices and will move.

ProductPriceCapsEPA+DHA/capCost per 1 g EPA+DHACaps for 1 gCaps for 4 g
Healthy Care Fish Oil 1000 mg$24.99400300 mg$0.213–413
Nature's Own Odourless 1000 mg$34.99400300 mg (not broken out)$0.293–413
Blackmores Odourless 1000 mg$22.49 (RRP $44.99)200300 mg$0.373–413
Blackmores Omega Triple$55.99 (RRP $87.99)150900 mg$0.41~1~4.5
Ethical Nutrients High Strength$43.97 (RRP $87.95)120654 mg$0.56~1.5~6

Workings: cost per gram = price ÷ (capsules × mg per capsule ÷ 1000).

Four things worth noticing:

1. The cheapest generic is the cheapest omega-3 by a wide margin — about 2.7× cheaper per gram than the premium concentrate. The premium brands sell convenience (fewer capsules), not cheaper omega-3.

2. Thirteen capsules a day is not a realistic regimen. This is the practical reason the 4 g cardiovascular dose is a prescription proposition and not a supermarket one, and the reason that almost nobody taking standard fish oil has ever taken a REDUCE-IT dose.

3. The discounting is extreme — $87.95 down to $43.97 in one case. Cost-per-gram comparisons are meaningless unless done on the same day.

4. The prescription option costs about the same as the capsules. Vazkepa (icosapent ethyl) was PBS-listed on 1 October 2024, item 14617H: dispensed price for maximum quantity $150.64, patient co-payment $25.00, 120 capsules with 5 repeats, Streamlined Authority. At 4 capsules a day that is a 30-day supply.

Restriction: established atherosclerotic cardiovascular disease with hypertriglyceridaemia — specifically coronary, cerebrovascular/carotid or peripheral arterial disease; stable statin therapy ≥12 weeks at target LDL (or statin-intolerant); LDL-C 1.0–2.6 mmol/L; fasting triglycerides 1.7–5.6 mmol/L.

At $0.21 per gram, 4 g/day of generic fish oil costs about $25 a month — almost exactly the $25 PBS co-payment for a regulated pharmaceutical with an outcome trial behind it, verified content and an AF warning on the label. For an eligible patient, the prescription is the better buy at the same price. Ask your GP whether you qualify.


Fish beats capsules, and tuna is not fish oil

EPA+DHA per 85 g cooked serve (NIH Office of Dietary Supplements):

FoodEPA+DHAServes for 1 g
Atlantic salmon, farmed1.83 g~0.55
Atlantic herring1.71 g~0.58
Atlantic salmon, wild1.57 g~0.64
Sardines, canned in tomato sauce1.19 g~0.84
Atlantic mackerel1.02 g~1.0
Oysters, wild0.53 g~1.9
Light tuna, canned in water0.19 g~5.3

One ~100 g serve of salmon, herring or mackerel — or one small tin of sardines — delivers roughly 1–2 g of EPA+DHA. That is as much as or more than the dose used in VITAL, ASCEND and VITAL-Rhythm, for about $2–3.

A tin of light tuna in water is nearly worthless for omega-3 — about 190 mg, so you would need five tins for a gram. "I eat tuna sandwiches" is a very common and very wrong substitute.

The Heart Foundation recommends 2–3 serves of fish (150–200 g total) per week, including oily fish, and states that "eating fish is the best, and the recommended, way to consume essential omega-3 nutrients for heart health." Supplements are positioned as a fallback. It recommends 1 g/day supplementation only as adjunctive therapy in heart failure. That document is dated 2015.

Be careful with the "fish beats supplements" argument, though. Observational cohorts consistently associate fish eating with lower cardiovascular mortality while supplement RCTs are null — but people who eat fish differ systematically from people who don't: higher income, less smoking, more exercise, better healthcare access. The cohort data cannot separate the fish from the fish-eater. The two defensible versions of the argument are substitution (oily fish displaces red and processed meat, and brings protein, selenium, iodine, vitamin D and a food matrix a capsule does not) and the negative one: supplements have been tested properly and don't deliver, so there is no reason to use a capsule to replicate a food whose benefit may not reside in the oil at all.


The claims that don't hold up

Depression

The optimistic meta-analysis: Liao Y, et al. Transl Psychiatry 2019;9:190. 26 RCTs, n=2,160. Overall SMD −0.28, p=0.004; EPA-pure −0.50; EPA-major (≥60% EPA) −1.03; DHA-predominant no benefit. I² = 75%.

The better one: Appleton KM, Voyias PD, Sallis HM, et al. Cochrane Database Syst Rev 2021;(11):CD004692. 35 studies, n=1,924. SMD −0.40 (95% CI −0.64 to −0.16), which the authors say is "unlikely to be clinically meaningful" — about 2.5 points on the 17-item Hamilton scale, below the 3.0-point minimal clinically important difference. Certainty: VERY LOW. The authors add that their own estimate "may be biased towards a positive finding."

The large clean trial: VITAL-DEP (Okereke OI, et al. JAMA 2021;326(23):2385–2394), n=18,353, 1 g/day, 5–7 years. Null for prevention of depression, with a slight increase in depressive symptoms. Okereke: "there is no reason for people in the general population to take omega-3 fish oil supplements solely for the purpose of preventing depression or for maintaining a positive mood."

The pattern — moderate pooled effects in small trials, null in the two largest, Cochrane rating its own positive estimate as probably biased upward — is the signature of small-study and publication bias. The EPA-ratio finding comes entirely from subgroup comparisons across heterogeneous small trials. Treat it as a hypothesis, not a dosing rule.

Pregnancy — and a Cochrane review overturned

Cochrane (Middleton P, Gomersall JC, Gould JF, et al. 2018;(11):CD003402), 70 RCTs, n=19,927, graded high quality:

  • Preterm <37 weeks: RR 0.89 (0.81–0.97)
  • Early preterm <34 weeks: RR 0.58 (0.44–0.77)
  • Prolonged gestation >42 weeks: RR 1.61 (1.11–2.33)
  • And "very few differences" for child cognition, IQ, vision, neurodevelopment, growth, language and behaviour

Then ORIP (Makrides M, Best K, Yelland L, et al. NEJM 2019;381(11):1035–1045) — n=5,544 pregnancies, South Australian-led, 900 mg/day n-3 from <20 weeks:

  • Early preterm delivery (<34 weeks): 2.2% vs 2.0%, adjusted RR 1.13 — NULL
  • No effect on most neonatal outcomes
  • Harm signal: infants in the omega-3 group were more likely to be large for gestational age

ORIP alone was larger than a quarter of that entire Cochrane evidence base, and it was null. A "high-quality" Cochrane finding overturned by the next big trial is worth sitting with.

The subgroup, and the Australian policy built on it. The ORIP benefit was confined to women with low baseline blood omega-3 carrying a singleton — about 20% of pregnant women (exploratory analysis, Simmonds LA, et al. BJOG 2020;127(8):975–981; we have not read the effect sizes and are not quoting them). This drove real policy: SAHMRI and SA Pathology launched an Omega-3 Test-and-Treat Program in 2021, screening more than 22,000 South Australian women, around 800 a month, with ~17% found to have low omega-3 and advised to supplement.

A positive subgroup in a null trial is a hypothesis, not a result. Treat the SA program as a reasonable policy bet built on exploratory analysis, not as proven. And the "infant brain development" claim on fish oil marketed to pregnant women is not supported by Cochrane at all.

Dry eye

DREAM (Asbell PA, Maguire MG, Pistilli M, et al. NEJM 2018;378(18):1681–1690). n=535, 3,000 mg/day (2,000 EPA + 1,000 DHA) vs refined olive oil, 12 months.

OSDI symptom score: −13.9 active vs −12.5 placebo; mean difference −1.9 (95% CI −5.0 to 1.1), p=0.21. Null.

Both groups improved substantially — a clean illustration of placebo and regression effects in symptom trials, and the reason uncontrolled testimonials about fish oil and dry eye are worth nothing.

Cognition and dementia

Cochrane (Burckhardt M, et al. 2016;(4):CD009002), 3 RCTs, n=632, mild-to-moderate Alzheimer's: no benefit on cognitive function, no effect on activities of daily living. "No convincing evidence for the efficacy of omega-3 PUFA supplements in the treatment of mild to moderate AD." Certainty high for most outcomes. On the prevention side, AREDS2 found no cognitive benefit.

Rheumatoid arthritis — the one weak positive

**Lee YH, Bae SC, Song GG. Arch Med Res 2012;43(5):356–362. 10 RCTs, 183 vs 187, dose ≥2.7 g/day**, ≥3 months.

  • NSAID consumption: SMD −0.518 (95% CI −0.915 to −0.121), p=0.011 — the only significant outcome
  • Tender joints: −0.214, p=0.128 — NS
  • Swollen joints: −0.170, p=0.241 — NS
  • Morning stiffness: −0.224, p=0.221 — NS
  • Physical function: 0.264, p=0.314 — NS

One significant outcome of five, in 370 people total, and that outcome is the softest and least blindable (self-reported pill-taking). The dose is ≥2.7 g/day ≈ 9 standard capsules. Fair summary: a plausible modest NSAID-sparing effect at high dose, not a disease-modifying treatment, and the trials are too small for confidence.


Oxidation: contested, and the real problem is a different one

You may have read that most fish oil is rancid. The picture is messier than that.

**Albert BB, Cameron-Smith D, Hofman PL, Cutfield WS. Sci Rep 2015;5:7928** tested 36 New Zealand supplements:

  • Only 3 of 32 (9%) met label EPA+DHA claims; mean actual content 68% of label
  • Against limits PV <5, AV <20, TOTOX <26: PV exceeded in 83%, AV in 25%, TOTOX in 50%. Only 8% met all three
  • Best-before date, cost and country of origin were all poor markers of quality

It was contradicted twice in the same market.

**Nichols PD, Dogan L, Sinclair A. Nutrients 2016;8(11):703 — only 10 products, EPA+DHA at 109% of label, PV 100% compliant. Funded by the Australian and New Zealand Omega-3 Centre, an industry body.**

**Bannenberg G, Mallon C, Edwards H, et al. Sci Rep 2017;7:1488 — 47 products, same market, same journal. 91% met label claims; PV 72% compliant, p-AV 86%, TOTOX 77%.** The authors raised four substantive methodological objections to Albert 2015:

  1. Inadequate sample handling after capsule isolation can oxidise oil during analysis
  2. p-AV gives false positives in flavoured oils — 25 of 47 products contained flavourings, and most Australian and New Zealand retail fish oil is flavoured or "odourless" (i.e. deodorised)
  3. Accredited laboratories in proficiency-testing schemes are more reliable
  4. **Albert reported 38% of products with a PV/p-AV ratio >1, indicating recent oxidation, versus 0% here** — consistent with degradation during analysis rather than on the shelf

Bannenberg and a co-author are associated with GOED, the omega-3 industry trade body. This is industry rebutting an academic finding. Both sides have skin in the game, and there is a published back-and-forth in Nutrients 2017 that we have not read.

The independent Australian study lands in the middle. Heller M, Gemming L, Tung C, Grant R. Int J Food Sci Nutr 2019;70(5):540–550, 26 Australian supplements: PV exceeded in 38%, pAV in 25%, TOTOX in 33%.

Fair synthesis: across four studies in this market, failure rates against voluntary limits run from 0% to 83%, with the independent Australian study at 25–38%. The honest headline is "somewhere between a quarter and a third of Australian fish oil products exceed voluntary industry oxidation limits, and the most alarming figure in the literature is methodologically disputed" — not "83% are rancid."

There is also no Australian consumer-advocacy lab testing. CHOICE has not conducted independent laboratory testing of omega-3 supplements; its article offers storage advice and 2013 pricing.

Does oxidised fish oil harm you?

Probably not, and nobody really knows. The New Zealand Ministry for Primary Industries commissioned a risk profile in April 2015, directly in response to the Albert controversy:

  • "The level of oxidation products in the fish oils on the NZ market is not high enough to pose a risk to consumers" at recommended doses
  • But "human and indeed animal data appear not extensive enough to serve as the basis for derivation of an acceptable daily intake"
  • Crucially: "industry standards are based on palatability" — the PV/AV/TOTOX limits exist to stop oil tasting rancid, not because exceeding them is known to be dangerous

GOED (industry, weigh accordingly) says two human trials found no oxidative-stress signs even at 8 g/day of highly oxidised fish oil.

Three things to take from this:

  1. "Useless" is better supported than "harmful." The content finding — products delivering around 68% of labelled EPA+DHA in the 2015 study — is a bigger practical problem than oxidation, because it means the dose you calculated is not the dose you took.
  2. There is no good human evidence that oxidised fish oil at retail levels causes harm, and the regulator that investigated said so. Anyone claiming otherwise is over-reading animal data.
  3. The limits are palatability thresholds, so your nose is a reasonable instrument — which is exactly why "odourless" and flavoured capsules are a problem. They defeat the only quality check a consumer has.

That last point sets up a real trade-off: the standard advice to buy flavoured or odourless capsules to avoid fishy aftertaste directly conflicts with buying unflavoured so you can detect rancidity. You have to pick.


Form: ethyl ester, triglyceride, krill, algae

The best review we found is Alijani S, et al., Progress in Lipid Research 2025;97, on EPA and DHA bioavailability in humans.

Acute bioavailability hierarchy: free fatty acid > phospholipid (krill) > re-esterified triglyceride > natural triglyceride > ethyl ester. Ethyl esters are hydrolysed by pancreatic lipase 10–50× more slowly than triglycerides, and a high-fat meal substantially improves their absorption.

But the decisive finding: significant differences in acute studies "often did not translate into long-term impacts in chronic supplementation studies." After weeks to months of daily dosing, differences between forms largely disappear and are "likely not clinically significant" at steady state — because the health effects come from tissue and membrane incorporation over time, not from absorption rate.

So the one piece of actionable advice that survives: if your product is an ethyl ester (most concentrates are), take it with a fatty meal, not on an empty stomach. Everything else about "superior absorption" is a price justification.

Krill

The newest and most favourable trial to krill is also unusually clean. **Loukil I, Vachon A, Çaku A, Plourde M. Am J Clin Nutr 2026. n=72, 12 weeks, funded by the Canadian Institutes of Health Research, authors report no conflicts**.

  • Krill arm received less EPA+DHA (1,099 mg vs 1,200 mg)
  • ΔEPA at 12 weeks: 6.88 vs 4.90 mg/dL, p=0.0001
  • ΔDHA: 4.22 vs 2.84 mg/dL, p=0.005
  • About a 1.5-fold higher rise in plasma omega-3 per dose

Krill genuinely does raise plasma omega-3 more per milligram. But the outcome is a blood biomarker, not a clinical one — no trial shows krill oil prevents anything fish oil doesn't — and the economics are brutal. Krill capsules typically deliver 50–120 mg EPA+DHA each versus 300 mg for a cheap fish oil capsule, at several times the price per capsule. A 1.5× bioavailability advantage cannot rescue a 5–10× cost-per-milligram disadvantage. (We did not price Australian krill products; if you want the exact figure, compute price ÷ total EPA+DHA on the shelf.)

Algae, and why flaxseed is not a substitute

ALA → EPA → DHA conversion in humans is "very limited, with reported rates of less than 15%" (NIH Office of Dietary Supplements), and conversion to DHA specifically is far lower than to EPA. Flaxseed and chia are not substitutes for EPA+DHA. That is the important consumer message and it is solid.

Algal oil is different: it supplies preformed DHA, and in newer products EPA, so it is mechanistically a genuine equivalent. The expectation in the literature is that algal DHA raises blood DHA comparably to fish oil, but we could not verify a head-to-head comparison and are flagging "equivalent" as probable rather than demonstrated. Note also that most algal products are DHA-dominant with little EPA, which matters if your reason for taking it is triglycerides or the (weak) depression literature, both of which point to EPA.


Safety

Bleeding — mostly a myth at supplement doses

**Begtrup KM, Krag AE, Hvas AM. Dan Med J 2017;64(5):A5366. 52 publications**, 32 in healthy subjects and 20 in surgical patients, doses from 270 mg to ≥6 g/day.

  • "Fish oil exposure in surgical patients did not increase bleeding or blood transfusions either during or after surgery"
  • Two studies showed reduced transfusion requirements in the fish oil arms
  • Conclusion: "discontinuation of fish oil supplements prior to surgery is not recommended"

Platelet-function changes were measurable in healthy subjects but did not translate into clinical bleeding.

OPERA (cardiac surgery, n=1,516, 28 centres, 1 g/day): BARC type 4/5 bleeding 5.5% vs 6.6%; transfusion units 1.61 vs 1.92, RR 0.83, p<0.001 — the fish oil arm needed fewer transfusions.

But the prescription label disagrees, and both deserve reporting. The US Vascepa label warns of increased bleeding: any bleeding 12% vs 10%, serious 3% vs 2%, with explicitly increased risk on concomitant aspirin, clopidogrel or warfarin. REDUCE-IT's own serious-bleeding figure was 2.7% vs 2.1%, p=0.06.

Reconciliation: at 1 g/day, perioperative discontinuation is not supported and "fish oil thins your blood" is largely unsupported. At 4 g/day in patients already on antiplatelets, there is a small real signal. The NIH notes that 2–15 g/day might increase bleeding time. It is dose- and context-dependent, not a flat debunk — and the decision to stop before surgery belongs to your surgeon, not to a wiki.

Upper limits

  • FDA: supplement labels should not recommend more than 2 g/day EPA+DHA
  • EFSA: long-term intakes up to ~5 g/day appear safe

Note the tension: EFSA's 5 g sits awkwardly beside the AF dose-response data showing risk rising above 1 g/day.

Gastrointestinal effects

From STRENGTH, the largest clean read: GI disorders 24.7% vs 14.7% at 4 g/day; diarrhoea 11.9% vs 4.9%. Real and common at high dose.

Fishy burps — no trial evidence at all

We searched specifically for randomised or controlled evidence on reducing aftertaste — enteric coating, freezing capsules, taking with food, splitting doses, refrigeration — and found only commercial and advice-column content. No trial. The standard advice set is mechanically plausible and completely untested. And as noted above, buying flavoured or odourless capsules to avoid the aftertaste conflicts with buying unflavoured so you can smell rancidity.

Heavy metals — fish oil is genuinely clean, and there's a reason

**Torović L, et al. Proceedings 2023;91(1):364 — 42 fish oil supplements**:

  • Mercury 0.001–0.0057 mg/kg, mean 0.0019 mg/kg
  • Far below the 0.1 mg/kg maximum for food supplements
  • "The risk from mercury in fish oil supplements was negligible, even for pregnant and nursing women" — exposure under 0.24% of the oral reference dose, conservatively treating all mercury as methylmercury

An earlier US study (Foran SE, et al. 2003) found mercury non-detectable to 10–12 µg/L across five OTC brands, comparable to basal human blood concentrations.

The mechanism is satisfying: methylmercury in fish binds to cysteine residues in muscle protein, not to lipid. When oil is pressed and refined out of fish tissue, the mercury stays behind with the protein; molecular distillation of concentrates removes more still. That is why fish oil carries essentially no mercury even when made from fish that do. (We are confident in this chemistry but did not land a clean peer-reviewed citation for the protein-binding step specifically.)

Don't over-extend it, though. Fat-soluble contaminants — dioxins and PCBs — behave oppositely: they do partition into oil, and are removed by refining and distillation rather than excluded by chemistry. "Fish oil has no mercury" is not "fish oil has no contaminants."


What to do

If you eat two or three serves of oily fish a week, you don't need a supplement. That is the Heart Foundation's position and the evidence supports it. One tin of sardines is a day's capsules.

If you don't eat fish and want the general-population target (250–500 mg EPA+DHA/day): one or two cheap standard capsules. Buy on cost per gram of EPA+DHA, not on brand — the generics win by roughly 2.7×.

Don't take fish oil to prevent a heart attack. High-certainty Cochrane evidence says it doesn't, at any supplement dose.

If you have very high triglycerides and established cardiovascular disease, talk to your GP about Vazkepa rather than buying thirteen capsules a day. It is PBS-listed at a $25 co-payment, has verified content, and carries the AF warning your bottle doesn't.

If you have a history of atrial fibrillation or you get palpitations, and you're taking more than about a gram a day, raise it with your doctor. This is the single most under-communicated fact about fish oil.

Don't buy it for depression, dry eye, memory, or your baby's brain. Those claims are not supported and in several cases have been specifically tested and failed.

Take ethyl-ester concentrates with a fatty meal. Skip krill unless you want to pay several times more per milligram for a biomarker advantage.

And if your capsules smell strongly rancid, bin them — not because we know oxidised oil harms you, but because the limits exist for palatability and your nose is the only test you own.


Frequently asked

Does fish oil prevent heart attacks? No, at the doses sold over the counter. A Cochrane review of 86 trials in 162,796 people found little or no effect on all-cause mortality or cardiovascular events, both at high certainty. The one positive trial, REDUCE-IT, used a prescription drug at 4 g/day in a selected high-risk population, and its mineral-oil placebo is disputed. A near-identical trial, STRENGTH, found nothing.

How much fish oil should I take? It depends on why. The Heart Foundation's general-population target is 250–500 mg of EPA+DHA a day, which is one or two standard capsules or a couple of fish meals a week. Every cardiovascular outcome trial used 840 mg to 4,000 mg — three to thirteen capsules. Read the EPA and DHA figures in the ingredients panel, not the "1000 mg" on the front.

Does fish oil cause atrial fibrillation? There is a real dose-dependent signal. A pooled analysis of 7 trials in 81,210 people found an overall hazard ratio of 1.25, rising to 1.49 above 1 g/day. At about 1 g/day the largest single trial (VITAL-Rhythm, n=25,119) found no significant increase. The EMA, UK MHRA and US FDA have all added atrial fibrillation to the prescription omega-3 label; we could find no such warning on Australian over-the-counter fish oil.

Is most fish oil rancid? The famous 2015 New Zealand study found 83% exceeding peroxide limits, but two later studies in the same market found far lower failure rates and raised substantive methodological objections — notably that the assay gives false positives in flavoured oils, and most retail fish oil is flavoured. The independent Australian study found 25–38% exceeding voluntary limits. There is no good human evidence that oxidised fish oil at retail levels causes harm; the New Zealand regulator investigated and said the levels pose no risk. The bigger practical problem is content: the 2015 study found products averaging 68% of their labelled EPA+DHA.

Is krill oil better than fish oil? It raises plasma omega-3 about 1.5× more per milligram in a clean independent trial. But krill capsules typically carry 50–120 mg of EPA+DHA each against 300 mg for a cheap fish oil capsule, at several times the price, so the cost per milligram is far worse. And no trial shows krill prevents anything fish oil doesn't.

Should I stop fish oil before surgery? A systematic review of 52 publications concluded that discontinuation before surgery is not recommended, and a trial in 1,516 cardiac surgery patients found the fish oil arm needed fewer transfusions. At prescription doses of 4 g/day alongside aspirin or warfarin there is a small real bleeding signal. Ask your surgeon — this is their call, not a general rule.

Can I get enough omega-3 from flaxseed? No. Humans convert plant ALA to EPA at rates under 15%, and to DHA far less. Algal oil is a genuine alternative because it supplies preformed DHA, though most algal products are DHA-dominant with little EPA.

Is there mercury in fish oil? Essentially none. Testing of 42 supplements found mercury far below the limit for food supplements, with consumer exposure under 0.25% of the reference dose. Methylmercury binds to fish muscle protein rather than to fat, so it stays behind when the oil is pressed and refined. Fat-soluble contaminants like dioxins and PCBs behave differently and are removed by refining rather than excluded by chemistry.


Last reviewed 1 October 2026. Manufacturer and industry funding is noted on every study where the source discloses it. Prices are Australian retail at the date of review and move constantly. Where we could not verify a figure, we have said so rather than printing it. This page is general information, not medical advice — and specifically, the atrial fibrillation and bleeding sections are reasons to talk to your doctor, not reasons to change a prescribed treatment on your own.