REDUCE-IT gave 8179 patients 4 g/day of pure EPA and cut major cardiovascular events 25%. STRENGTH gave 13,078 patients 4 g/day of mixed EPA and DHA and cut nothing. Same dose, opposite result.
EPA is eicosapentaenoic acid, 20 carbons and 5 double bonds. DHA is docosahexaenoic acid, 22 carbons and 6 double bonds. Two carbons and one double bond apart. On paper they are near-twins. In the body they behave like different compounds.
The difference is where they go and what they become. DHA is the structural fatty acid. It makes up over 50% of the phospholipid fatty acid side chains in photoreceptor outer segments[38], and it concentrates in neural membranes. EPA does not accumulate that way. It sits in circulating phospholipids and cholesteryl esters and turns over faster. Supplementing DHA raises EPA slightly through retroconversion. Supplementing EPA does not raise DHA[19].
That divergence shows up in trials as opposite lipid effects at matched doses. Against placebo, DHA raised LDL cholesterol by 7.23 mg/dL, 95% CI 3.98 to 10.5, while EPA produced a non-significant reduction[16]. In metabolic syndrome the split is starker: EPA lowered LDL-C by 0.13 mmol/L while DHA raised it by 0.26 mmol/L[20]. In a 59-person head-to-head, DHA improved forearm blood flow to acetylcholine, P=0.040, and EPA did not[14].
So the ratio question is not marketing invention. Two fatty acids, one bottle, different jobs. The failure mode is buying a ratio for the wrong job.
Here is what the landmark trials actually put in the capsule. Note how rarely the tested ratio resembles anything on a retail shelf.
Swipe the chart sideways
Start with the two trials that make the whole argument.
REDUCE-IT. 8179 statin-treated patients, 70.7% with established cardiovascular disease, fasting triglycerides 135 to 499 mg/dL, LDL-C 41 to 100 mg/dL. Icosapent ethyl 2 g twice daily, which is 4 g/day of highly purified EPA ethyl ester with zero DHA. Median follow-up 4.9 years. The primary composite hit 17.2% against 22.0%, HR 0.75, 95% CI 0.68 to 0.83, P<0.001. Cardiovascular death, myocardial infarction and stroke: 11.2% against 14.8%, HR 0.74. Cardiovascular death alone: 4.3% against 5.2%, HR 0.80, P=0.03[1]. Counting all events rather than first events, the reduction reaches 30%[2].
STRENGTH. 13,078 statin-treated high-risk patients across 675 sites in 22 countries. Omega-3 carboxylic acid, a mixed EPA and DHA preparation, 4 g/day. Primary composite: 785 events, 12.0%, against 795, 12.2%. HR 0.99, 95% CI 0.90 to 1.09, P=0.84. The trial was halted for futility. Gastrointestinal adverse events ran 24.7% against 14.7%[3].
Same total dose. Same patient type. Same era. One difference in the active arm, DHA, and one difference in the control arm, mineral oil against corn oil. That second difference is why cardiologists still argue[4].
The honest reading: purified EPA has now shown benefit in two independent trials, at 4 g/day and at 1800 mg/day, in two very different populations. Mixed EPA+DHA has shown benefit on myocardial infarction across pooled data, RR 0.87, 95% CI 0.80 to 0.96, with an explicitly dose-dependent effect for cardiovascular events and myocardial infarction[8]. What mixed preparations have not done is beat a composite endpoint at 4 g/day.
Observational data refuse to pick a winner. In 6501 women followed a median 14.9 years, all-cause mortality hazard ratios per standard deviation of red-cell content were 0.89 for EPA and 0.93 for DHA[12].
One safety number belongs in this section. REDUCE-IT reported hospitalisation for atrial fibrillation or flutter at 3.1% against 2.1%, P=0.004, and serious bleeding at 2.7% against 2.1%, P=0.06[1]. OMEMI showed the same directional signal for new atrial fibrillation[6]. High-dose omega-3 is not risk-free.
This is the cleanest teachable trade-off in the whole topic, and the numbers come from trials that gave one purified fatty acid at a time.
At matched 4 g/day doses over 6 weeks in 59 overweight mildly hyperlipidemic men, triacylglycerols fell 0.45 mmol/L, about 20%, with DHA, P=0.003, and 0.37 mmol/L, about 18%, with EPA, P=0.012. Neither changed total cholesterol. DHA raised HDL2-C by roughly 29%, P=0.004, and nudged LDL-C up. EPA lowered HDL3-C by 6.7%, P=0.032[13]. In 51 patients with type 2 diabetes and treated hypertension, triacylglycerols fell 19% on EPA and 15% on DHA, P=0.022, with no effect on HbA1c, insulin or insulin sensitivity in either arm[15].
Swipe the chart sideways
Two corrections keep this from becoming a simple rule.
First, the LDL particles get bigger. In a 154-person crossover at 2.7 g/day for 10 weeks, DHA raised LDL-C 3.3% against EPA, P=0.038, but also increased mean LDL particle size by 0.7 angstrom, P<0.001, and cut the proportion of small LDL by 3.2%, P<0.01. Both arms lowered PCSK9 similarly, 18.2% against 25.0%[18]. A larger LDL population at a slightly higher cholesterol concentration is not obviously worse.
Second, baseline triglycerides flip the hierarchy. In 49 normolipidemic subjects at 3 g/day of ethyl esters, EPA and a fish oil concentrate lowered triglycerides and VLDL-cholesterol, P<0.01, and DHA did nothing to triglycerides, VLDL or LDL at all[19]. So the DHA triglyceride advantage is a hyperlipidemic phenomenon, not a universal one.
The American Heart Association's own position is deliberately unexcited about ratio. It defines hypertriglyceridemia as 200 to 499 mg/dL, very high as 500 mg/dL or above, and pharmacological dosing as over 3 g/day total EPA+DHA. At 4 g/day both EPA+DHA agents and EPA-only agents reduce triglycerides by 30% or more, and the advisory calls them roughly comparable for triglyceride lowering. The difference it flags is LDL: EPA-only did not raise LDL cholesterol in very high triglycerides, whereas EPA+DHA agents produced concurrent LDL-C increases in that population[21].
The zero-DHA proof is MARINE. 229 patients with fasting triglycerides between 500 and 2000 mg/dL received an EPA ethyl ester that was at least 96% EPA with no DHA. Placebo-corrected median triglyceride change at 12 weeks was -33.1% at 4 g/day, P<0.0001[22]. DHA is not required to cut triglycerides by a third.
Dose, though, is required. In a 25-person crossover, apoB and apoC-III fell significantly at 3.4 g/day, P<0.01, with VLDL-C down at P<0.005, while 0.85 g/day did nothing[23]. Blood pressure follows the same shape: across 71 trials and 4973 people, systolic pressure fell 2.61 mm Hg and diastolic 1.64 mm Hg at 2 g/day, with the optimum combined intake placed between 2 and 3 g/day[24]. Within that, EPA edges systolic, -2.6 mmHg, and DHA edges diastolic, -3.1 mmHg in dyslipidemia[25].
Swipe the chart sideways
DHA is the brain's structural omega-3. It is also over 50% of the phospholipid fatty acid side chains in photoreceptor outer segments[38]. That biology is real. It does not transfer to supplement outcomes, and the reason is worth understanding.
The one clear win. MIDAS randomized 485 healthy adults aged 55 and over with age-related cognitive decline to 900 mg/day DHA or placebo for 24 weeks across 19 US sites. Paired associate learning six-pattern errors fell, difference score -1.63, 95% CI -3.1 to -0.14, P=0.03. Immediate and delayed verbal recognition memory improved, P<0.02. Plasma DHA doubled and correlated with the learning improvement, P<0.02. Working memory and executive function were unchanged[26]. That is 900 mg of DHA with no added EPA, a ratio no mainstream fish oil sells.
The nulls. AREDS2's cognitive ancillary study tested 1 g/day of a 650 mg EPA plus 350 mg DHA blend in 3741 consenting participants over 5 years. Yearly change in the composite cognitive score differed by -0.03, 99% CI -0.20 to 0.13, P=0.63[27]. In established Alzheimer disease, 2 g/day of algal DHA for 18 months in 402 patients moved nothing: ADAS-cog rose 7.98 points against 8.27 on placebo, P=0.41, and brain volume declined 1.32% per year against 1.29%, P=0.79[28]. In mild cognitive impairment, 1491 mg DHA plus 351 mg EPA daily for 12 months in 72 older adults produced no evidence of a treatment effect on cognitive measures, though systolic pressure improved, P=0.03, and APOE e4 carriers showed treatment interactions favouring DHA for depression, P=0.04, and anxiety, P=0.02[29].
Read those four trials together and the pattern is about timing and endpoint, not ratio. DHA-dominant dosing helped a memory task in healthy older adults with early decline. It did nothing once disease was established, at any dose up to 2 g/day.
The mechanistic reason. Fatty acid profiling of ocular tissue from 27 human donors found that DHA in the neural retina was poorly associated with its dietary intake, unlike linoleic acid[39]. Retinal DHA is homeostatically defended. AREDS2's primary macular degeneration result matches: progression to advanced disease with DHA+EPA gave HR 0.97, 98.7% CI 0.82 to 1.16, P=0.70, in 4203 participants over 5 years[37]. Being made of a nutrient does not mean responding to more of it.
Mood flips the ratio. Depressive symptoms are the one central-nervous-system endpoint where the ratio evidence is clean, and it points at EPA. Across 26 double-blind trials and 2160 participants the overall effect was SMD -0.28, P=0.004. Pure-EPA formulations at 1 g/day or less gave SMD -0.50, P=0.003. EPA-major formulations, 60% EPA or more, at 1 g/day or less gave SMD -1.03, P=0.03. DHA-pure formulations did not exhibit efficacy[30]. An earlier 28-study pooling reached the same threshold: preparations over 50% EPA were efficacious, pure DHA and DHA-dominant preparations were not[31]. That 50 to 60% EPA rule is the most replicated ratio finding in the literature.
Ratio is the wrong variable here. The pooled head-to-head answer is a statistical tie.
A network meta-analysis compared DHA against EPA directly in 5 trials, N=411, and indirectly across 20 trials, N=1231. Pairwise mean differences: CRP 0.14 mg/L, 95% CI -0.57 to 0.85, with I-squared 61%. IL-6 0.10 pg/mL, 95% CI -0.15 to 0.34. TNF-alpha -0.10 pg/mL, 95% CI -0.37 to 0.18[35]. A separate 20-trial pooling put EPA at -0.56 mg/L on CRP and DHA at -0.50 mg/L[25]. Indistinguishable.
What does separate results is dose, and the picture is less flattering than supplement copy suggests.
Swipe the chart sideways
Across 68 randomized trials and 4601 subjects, marine omega-3 did lower TNF-alpha, IL-6 and CRP in chronic non-autoimmune disease, chronic autoimmune disease and healthy subjects. Two qualifications came with it: a significant negative linear relationship between supplementation duration and effect size, meaning the effect fades with time on treatment, and weaker effects when BMI exceeded 30 kg/m squared[34].
One widely quoted number deserves correction. A frequently cited 43% hsCRP reduction from 300 mg EPA plus 200 mg DHA daily for 8 weeks in 64 obese hypertensive or diabetic adults was a within-group change, hsCRP 14.78 to 8.49 mg/L, P<0.001. The control group also improved fasting glucose and triglycerides, and the trial's own conclusion was that the omega-3 arm did not reach clinical significance for any variable in the between-group comparison[33]. Within-group numbers are not treatment effects.
Practical reading: to move hsCRP, dose above 3 g/day and pay nothing extra for a ratio. Above BMI 30, or after years on treatment, expect less[34].
Pregnancy is the only endpoint where a regulator explicitly specifies DHA over EPA.
European dietary reference values set an adequate intake of 250 mg/day of EPA+DHA for adults, 250 mg/day for children aged 1 to 17, and 100 mg/day for infants aged 7 to 11 months. For pregnancy and lactation the figure is 250 mg/day EPA+DHA plus an additional 100 to 200 mg/day of DHA on top[59]. The long-standing international consensus floor is at least 200 mg DHA/day, with up to 1 g/day DHA and up to 2.7 g/day of total long-chain omega-3 used in trials without significant adverse effects[44]. A 2024 scoping review of 14 guideline documents confirms the 250 mg/day consensus and records one guideline advising 800 mg/day DHA with 100 mg/day EPA in women of low omega-3 status, an 8:1 DHA:EPA ratio[43].
What the trials show is more sober than the guidance implies. DOMInO randomized 2399 women before 21 weeks' gestation to DHA-rich fish oil delivering 800 mg/day DHA or matched vegetable oil until birth, with 96.7% completing. High maternal depressive symptoms in the first 6 months postpartum ran 9.67% against 11.19%, a non-significant difference, and child cognitive and language scores at 18 months did not differ[40]. A 957-woman trial of 400 mg/day algal DHA from 20 weeks' gestation through 6 months postpartum tracked 878 mother-infant pairs to 12 months[41]. In preterm infants born under 33 weeks, tripling feed DHA from about 0.3% to about 1% of total fatty acids produced a full-scale IQ of 98.3 at age 7 with no significant difference on the primary outcome[42].
Three numbers define the freshness standard. Peroxide value at or below 5 meq/kg, anisidine value at or below 20, and TOTOX at or below 26, where TOTOX equals twice the peroxide value plus the anisidine value[61]. Those are the voluntary industry limits, not law.
Market compliance depends entirely on which survey you read.
Swipe the chart sideways
The variable that predicts failure is not price or brand. It is flavoring. Among 72 US supplements sampled between 2014 and 2020, 68% of the 34 flavored products exceeded the TOTOX limit against 13% of the 38 unflavored ones. On peroxide value the split was 65% against 32%. Anisidine value failed in 62% of flavored products[47]. Flavoring masks the taste that would otherwise tell you the oil has turned.
Price buys nothing here. In the Australian survey, four premium-marketed supplements delivered 165 mg of fish oil per capsule against 577 mg for standard products, P=0.007, at $2.97 per gram against $0.39, P<0.001, with no difference in any oxidative marker[48]. That is 7.6 times the price for identical freshness and less oil.
Independent retail testing across 60 brands from 2014 to 2018 reported an average oxidation score of 24.4 against the 26 limit, with over 10% of products spoiled and nearly half close to the threshold. Those figures are secondary reporting of commercial testing rather than peer-reviewed data, and the most extreme individual result quoted should be treated as unverified[64]. A 2026 retail review of 10 tested supplements plus 18 approved through a voluntary certification programme found one product failed freshness testing, with EPA+DHA concentration in the oils ranging from about 33% to 85%[63].
Two bottles can declare the same 1000 mg of EPA+DHA and deliver measurably different amounts into your blood. The variable is how the fatty acids are attached.
Swipe the chart sideways
The blood-level advantage persists over months. In 150 volunteers taking 1.01 g EPA plus 0.67 g DHA daily for 6 months, the omega-3 index rose 186% against 161% at 3 months and 197% against 171% at 6 months, rTG against ethyl ester, P<0.01[49]. In the dyslipidemic statin-treated arm of the same programme, fasting triacylglycerols fell significantly from baseline at 3 and 6 months in the rTG group but not in the ethyl ester group, although the head-to-head difference between the two omega-3 groups was not itself significant[50].
Now the counterweight, because the form argument is routinely oversold. In 66 healthy adults over 28 days at 1.3 g/day EPA+DHA, total plasma EPA+DHA at 672 hours reached 90.9 micrograms/mL for fish oil ethyl ester, 108 for fish oil triglyceride and 118.5 for krill, P=0.052. The between-group bioavailability difference was under 24% and red-cell DHA+EPA gave P=0.19[52]. A 12-man acute crossover at a single 1680 mg dose put krill's 72-hour incorporation AUC highest at 80.03 percent-hours against 59.78 for rTG and 47.53 for ethyl ester, but standard deviations were large and no comparison reached significance[51]. A 2025 review sets the isolated-form ranking as free fatty acid, then phospholipid, then re-esterified triglyceride, then unmodified triglyceride, then ethyl ester, while concluding that significant differences in acute studies often did not translate into long-term impacts in chronic supplementation[53].
Physical delivery form is a second lever. In a rat lymph-duct model, emulsifying marine phospholipids raised total intestinal fatty acid absorption 84% against bulk oil[54]. Animal mechanistic data, not a dosing recommendation.
| Form | Relative bioavailability | Acute evidence | Chronic evidence | Where you find it |
|---|---|---|---|---|
| Re-esterified triglyceride, rTG | 134% of natural fish oil | Best absorbed form tested head to head[60] | Index rose 197% at 6 months against 171% for ethyl ester[49] | Most premium fish oils. Costs more to make |
| Natural triglyceride, TG | 100%, the reference | Fish body oil and cod liver oil grouped as the reference[60] | Lower EPA+DHA concentration per gram of oil | Cod liver oil and minimally processed fish oils |
| Free fatty acid, NEFA | 86% | Not statistically different from ethyl ester, P=0.29[60] | Ranked top of the isolated forms in a 2025 review[53] | Rare on shelves. Poor shelf stability |
| Ethyl ester, EE | 76% | Lowest of five forms, P=0.0001 against rTG[60] | Failed to lower triglycerides where rTG succeeded[50] | Most concentrated cheap fish oil, and REDUCE-IT's drug |
| Phospholipid, krill | No significant advantage at 4 weeks | AUC over 72 h favoured krill but P was not significant[51] | At 28 days all three forms sat within 24%, P=0.052[52] | Costs $1.38 to $2.60 per 500 mg EPA+DHA[63] |
One irony worth stating. REDUCE-IT's icosapent ethyl is an ethyl ester, the least bioavailable form in the comparison[60], and it produced the largest cardiovascular effect ever recorded for an omega-3[1]. Dose and purity outranked form there. At supplement doses, where you are trying to reach an omega-3 index of 8% rather than saturate a drug target, form matters more[11].
Algal DHA is not a compromise on DHA. It is bioequivalent.
In 32 healthy adults aged 20 to 65 over 2 weeks, 600 mg DHA/day from algal-oil capsules and the same DHA from assayed portions of cooked salmon both raised plasma phospholipid DHA by about 80% and erythrocyte DHA by about 25%, with similar changes between groups. The conclusion was explicit: as measured by delivery of DHA to plasma and erythrocytes, fish and algal-oil capsules were equivalent[55]. A 28-day study across two algal oils at 200, 600 and 1000 mg DHA/day found equivalent plasma phospholipid and erythrocyte DHA between strains, a linear dose response of 1.17, 2.28 and 3.03 g per 100 g fatty acid, and equivalent delivery from DHA-fortified snack bars[56].
The biomarker result is the strongest single argument. 104 healthy vegetarians took 0.94 g/day of DHA from microalgae oil or olive oil for 8 weeks. Red-cell total lipid DHA reached 7.9 against 4.4 weight percent. The omega-3 index rose from 4.8 to 8.4 weight percent, and 69% of supplemented subjects exceeded 8 weight percent against no subject on placebo. EPA rose only slightly[57].
Swipe the chart sideways
The caveat, plainly. Algae oil fixes DHA status and barely moves EPA[57]. EPA carries the depression signal, SMD -1.03 for EPA-major formulations[30], and the larger cardiovascular mortality effect, HR 0.79 against 0.92[9]. Some algal oils now include EPA at roughly 0.5:1 EPA:DHA, but no vegan product on our shelves reaches an EPA-forward ratio.
Cost is competitive. Retail testing puts the price of 500 mg of EPA+DHA at 16 to 56 cents for fish oil, 48 to 61 cents for algal oil and $1.38 to $2.60 for krill oil[63]. Algae sits at roughly the top of the fish oil range and well below krill.
We hold 37 omega-3 products. Their EPA:DHA ratios span 0.32:1 to 3.00:1, a roughly 9-fold spread. Fish oils cluster between 1.3:1 and 2.6:1 EPA:DHA. Algae products cluster at 0.5:1, which is 2:1 DHA:EPA, the mirror image.
| Product | EPA mg | DHA mg | Ratio EPA:DHA | Total mg | $/month | Form and testing |
|---|---|---|---|---|---|---|
| Triple Strength Omega-3 Fish Oil | 1500 | 570 | 2070 | 2.63:1 | 37.33 | rTG, IFOS and Labdoor tested |
| Ultimate Omega 2X | 1125 | 875 | 2000 | 1.29:1 | 52.95 | rTG, tested by independent lab |
| Omega-3 Fish Oil, VivoMega | 1100 | 800 | 1900 | 1.38:1 | 26.66 | Triglyceride form |
| O3 Ultra Pure Fish Oil | 1250 | 500 | 1750 | 2.50:1 | 42.75 | IFOS TOTOX 6.86, lowest published |
| Lion Heart Pure Omega 3 | 1220 | 510 | 1730 | 2.39:1 | 40.00 | rTG, IFOS TOTOX 11.65 |
| OmegAvail Hi-Po | 800 | 800 | 1600 | 1.00:1 | 43.49 | Triglyceride form |
| Elite Omega-3 Gems | 800 | 600 | 1400 | 1.33:1 | 28.28 | Tested by independent lab |
| Omega 3 Fish Oil Triple Strength | 800 | 600 | 1400 | 1.33:1 | 9.32 | Cheapest per 1000 mg in the set |
| The Very Finest Fish Oil, per teaspoon | 800 | 500 | 1300 | 1.60:1 | 14.03 | Liquid TG, IFOS 5-star |
| Extra Virgin Cod Liver Oil, per 5 ml | 510 | 700 | 1210 | 0.73:1 | 47.99 | Raw, DHA-dominant fish oil |
| Peak EPA | 750 | 250 | 1000 | 3.00:1 | 14.99 | Most EPA-forward in the catalogue. Ethyl ester |
| UnoCardio 1000 + Vitamin D | 665 | 445 | 1110 | 1.49:1 | 24.98 | rTG, IFOS TOTOX 10.82 |
| Omega-3, algae, life'sOMEGA | 270 | 540 | 810 | 0.50:1 | 44.10 | Vegan TG |
| Opti3 Omega-3 | 300 | 500 | 800 | 0.60:1 | 24.95 | Vegan, Schizochytrium, plus D3 |
| Algae Omega | 195 | 390 | 585 | 0.50:1 | 34.49 | Vegan |
| Ovega-3 500 mg | 90 | 280 | 370 | 0.32:1 | 22.97 | Most DHA-forward in the catalogue |
| Vegan Omega 3 DHA + DPA | 0 | 225 | 225 | DHA only | 12.99 | Zero EPA, plus 35 mg DPA |
EPA and DHA figures are per serving as recorded in our product data. Ratio is EPA divided by DHA, so values below 1 are DHA-dominant.
Four findings matter more than the ranking.
No product in the catalogue matches REDUCE-IT. Nothing is pure EPA with zero DHA. The closest is a 3.00:1 product at 750 mg EPA and 250 mg DHA, and it is an ethyl ester, the least bioavailable form in the head-to-head comparison[60]. Reaching 4 g/day of EPA from it means 5.3 servings per day. The trial dose is a prescription product, not a supplement purchase.
No product matches MIDAS either. The positive memory trial used 900 mg DHA with no added EPA[26]. Our most DHA-forward products are a DHA-only vegan capsule at 225 mg DHA and a 0.32:1 algae oil at 90 mg EPA and 280 mg DHA. Both are far below 900 mg DHA per serving. Three to four servings of the algae oil gets you there, at a ratio close enough.
Cost per delivered milligram varies about 7.7-fold. The cheapest triple-strength fish oil costs roughly $6.66 per month per 1000 mg/day of EPA+DHA. The most expensive comparable product costs roughly $51.18. Same fatty acids. Both tested by independent lab.
Published TOTOX is rare. Only 5 products in the whole category carry a public batch report with an actual oxidation number: TOTOX values of 6.86, 7.72, 10.82 and 11.65, and a peroxide value of 3.51. All sit comfortably under the limit of 26[61]. Given that 68% of flavored US products exceeded that limit[47], a published number is worth more than a claim of third-party testing.
Two data-quality flags. Krill products report only combined EPA+DHA, so their ratios cannot be computed. One low-cost product's label data names both an ethyl ester and a triglyceride blend, which we have not resolved.
Ranked by our six-pillar overall score. See the full ranking: Best Omega-3 Supplements 2026. No brand pays for placement.
Six goals. Each card gives the target ratio, the dose the trial used, and the trial that produced the number.
Target ratio: as close to pure EPA as you can buy
Dose: 4 g/day EPA
REDUCE-IT enrolled 8179 statin-treated patients on 4 g/day icosapent ethyl and hit MACE HR 0.75, 95% CI 0.68 to 0.83. Atrial fibrillation hospitalisation rose 3.1% against 2.1%.
Strongest trial evidenceRead REDUCE-ITTarget ratio: 2:1 EPA:DHA or higher
Dose: 3 to 4 g/day EPA+DHA
Pure EPA at 4 g/day cut triglycerides 33.1% placebo-corrected with no LDL rise. DHA alone raised LDL-C by 7.23 mg/dL, 95% CI 3.98 to 10.5.
Strong evidenceRead MARINETarget ratio: 1.5:1 EPA:DHA minimum, higher is better
Dose: 1 g/day or less
Across 26 trials and 2160 participants, EPA-major formulations at 1 g/day or less gave SMD -1.03, P=0.03. DHA-pure formulations showed no efficacy.
Consistent meta-analytic signalRead the meta-analysisTarget ratio: DHA-only or 4:1 DHA:EPA
Dose: 900 mg DHA/day
MIDAS randomised 485 adults over 55 to 900 mg/day DHA for 24 weeks. Paired associate learning errors fell by 1.63, 95% CI -3.1 to -0.14, P=0.03. Executive function was null.
One positive trial, several nullsRead MIDASTarget ratio: 0.5:1 EPA:DHA, algae oil default
Dose: 250 mg EPA+DHA plus 100 to 200 mg DHA
EFSA sets 250 mg/day EPA+DHA for adults with an added 100 to 200 mg DHA in pregnancy. One guideline goes to 800 mg DHA and 100 mg EPA in women of low omega-3 status.
Regulator-specifiedRead the guideline reviewTarget ratio: any
Dose: 1.7 g/day EPA+DHA as rTG for 6 months
At an identical 1.68 g/day, re-esterified triglyceride raised the index 197% against 171% for ethyl ester, P<0.01, in 150 volunteers over 6 months.
Direct biomarker dataRead the 6-month trialDoses are stated as the trials stated them. Two of these regimens, 4 g/day EPA for cardiovascular events and 3 to 4 g/day for triglycerides, are pharmacological doses with an atrial fibrillation signal[1][6] and belong in a conversation with a clinician, not a shopping cart.
| Endpoint | Target ratio | Dose | Best evidence | The qualification |
|---|---|---|---|---|
| Major cardiovascular events | EPA-forward, up to 100% EPA | 4 g/day | REDUCE-IT: HR 0.75, 95% CI 0.68 to 0.83, N=8179[1] | STRENGTH gave the same 4 g/day as a mixture and returned HR 0.99[3] |
| Cardiovascular mortality | EPA-forward | 1 to 4 g/day | Purified EPA HR 0.79 against mixed HR 0.92, 127,771 patients[9] | Both work. EPA roughly doubles the effect size |
| Myocardial infarction | Ratio not decisive | 1 g/day and up | VITAL total MI HR 0.72, 95% CI 0.59 to 0.90[5] | Pooled MI RR 0.87 and dose-dependent[8] |
| Triglycerides, TG 200 to 499 mg/dL | Either. DHA slightly ahead | over 3 g/day | Direct comparisons: TG -22.4% DHA against -15.6% EPA[17] | 0.85 g/day did nothing to lipids at all[32] |
| Triglycerides with high LDL-C | EPA-forward or pure EPA | 4 g/day | Pure EPA cut TG 33.1% with no LDL rise[22][21] | DHA alone raised LDL-C 7.23 mg/dL against placebo[16] |
| Blood pressure | Either | 2 to 3 g/day | SBP -2.61 mm Hg, DBP -1.64 mm Hg at 2 g/day[24] | EPA leads on systolic, DHA on diastolic[25] |
| Depressive symptoms | 60% EPA or more | 1 g/day or less | EPA-major at 1 g/day or less: SMD -1.03, P=0.03[30] | DHA-pure formulations showed no efficacy[30][31] |
| Memory in healthy older adults | DHA-dominant | 900 mg DHA/day | MIDAS: PAL error difference -1.63, P=0.03, N=485[26] | No product on the shelf delivers 900 mg DHA with no EPA |
| Established Alzheimer disease | No ratio works | 2 g DHA/day tested | ADAS-cog P=0.41, brain atrophy P=0.79, 18 months[28] | Timing beats ratio. Do not buy fish oil for this |
| Inflammation markers | Irrelevant | over 3 g/day | DHA against EPA on CRP: 0.14 mg/L, 95% CI -0.57 to 0.85[35] | Pure EPA at 4 g/day cut hsCRP 23 to 40%[36] |
| Age-related macular degeneration | No ratio works | 1 g/day tested | AREDS2: HR 0.97, 98.7% CI 0.82 to 1.16, 5 years[37] | Retinal DHA is defended against dietary change[39] |
| Pregnancy | DHA-forward, up to 8:1 DHA:EPA | 250 mg base plus 100 to 200 mg DHA | EFSA sets the only regulator-specified DHA add-on[59] | Consensus floor is 200 mg DHA/day[44] |
Step 1, name the endpoint. Cardiovascular events, triglycerides, mood, memory, pregnancy, or index repletion. If you cannot name it, buy the cheapest unflavored third-party-tested triglyceride-form fish oil at 1 g/day EPA+DHA and stop reading.
Step 2, set the ratio from the endpoint. EPA-forward for cardiovascular events, lipids with high LDL-C, and mood. DHA-forward for pregnancy and memory. Ratio-agnostic for blood pressure, inflammation markers and index repletion.
Step 3, set the dose from the trial, then check the form. Nothing below roughly 1 g/day has moved a hard endpoint[11][23]. Between two products at the same declared EPA+DHA, the re-esterified triglyceride delivers up to 1.7 times what the ethyl ester does[60].
There is no single best ratio. For major cardiovascular events the only trial that succeeded used 4 g/day of pure EPA with zero DHA and returned HR 0.75. For memory in healthy older adults the only positive trial used 900 mg/day of DHA with no added EPA. For pregnancy, European reference values add 100 to 200 mg of DHA specifically. Pick the endpoint first.
Yes, modestly. Against placebo, DHA raised LDL by 7.23 mg/dL, 95% CI 3.98 to 10.5, while EPA produced a non-significant reduction. In direct head-to-head comparisons the net difference is about 3.3 percentage points. The rise came with larger LDL particles, plus 0.7 angstrom, and 3.2% fewer small LDL particles, so the atherogenic meaning is not settled.
The European adequate intake for adults is 250 mg/day of EPA plus DHA. Trials that moved hard endpoints used 1000 to 4000 mg/day. Blood pressure optimum sits at 2 to 3 g/day. Atherogenic lipoproteins did not budge at 0.85 g/day but did at 3.4 g/day. The gap between adequate and therapeutic is roughly 16-fold.
It depends on the flavoring. Among 72 US supplements, 68% of the 34 flavored products exceeded the TOTOX limit of 26 against 13% of the 38 unflavored ones. Two surveys of the same national market disagreed, with 50% and 23% TOTOX failure rates four years apart. Two trials feeding 8 g/day of deliberately over-oxidized fish oil for 7 weeks found no oxidative-stress harm, so treat rancidity as a potency and palatability problem.
For DHA, yes. 600 mg/day of algal DHA was bioequivalent to the same DHA from cooked salmon, with plasma phospholipid DHA up about 80% in both arms. Algal DHA at 0.94 g/day took vegetarians from an omega-3 index of 4.8% to 8.4% in 8 weeks. What algae oil does not fix is EPA status, and EPA carries the depression and cardiovascular mortality signal.
Triglyceride form, if the price gap is small. Re-esterified triglyceride delivered 134% of natural fish oil against 76% for ethyl ester in a 72-person comparison at 3.3 g/day. Over 6 months the omega-3 index rose 197% on rTG against 171% on ethyl ester. At 4 weeks the difference shrinks to under 24% and loses significance, so the form advantage is real but smaller than most marketing implies.
Not on the data. At 28 days, krill, triglyceride and ethyl ester landed within 24% of each other, P=0.052. Krill costs $1.38 to $2.60 per 500 mg of EPA+DHA against 16 to 56 cents for fish oil.
Corrections and source queries: [email protected].