Fatty Fish

What it is

Fatty fish, such as salmon, sardines and mackerel, contain more long-chain omega-3 fats than lean fish, so studies often compare the two groups [1]. Canned light tuna was the least expensive option in one review of a US fish assessment [2]. Frying changes the picture: fried fish was linked to higher heart risk, while non-fried fish was linked to lower risk [3].

How it might work

Fish supplies long-chain omega-3 fats (EPA and DHA). These are proposed to have anti-inflammatory effects, to reduce platelet clumping and to regress atherosclerotic plaques [4]. A 2012 review suggested that the stroke benefit of fish may also come from the wide range of other nutrients fish provides [5].

What the research says

Heart disease. The largest reviews of cohort studies link fish eating to fewer heart disease events and deaths. A 2020 meta-analysis of 40 cohort studies found that the highest fish intake was linked to a lower risk of coronary heart disease (CHD) events (RR 0.91) and CHD deaths (RR 0.85) [6]. A 2022 review found that fatty fish, but not lean fish, was linked to lower CHD events (RR 0.92) and CHD deaths (RR 0.83) [1]. A 2019 food-group review found fish linked to a 12% lower CHD risk (RR 0.88) [7].

Heart attack. A 2014 meta-analysis of 408,305 people found that eating fish four or more times a week was linked to a 21% lower risk of acute coronary syndrome (RR 0.79); each extra 100 g of fish a week was linked to 5% lower risk (RR 0.95) [8]. A 2021 review of 24 studies found that non-fried fish was probably linked to fewer heart attacks, while fried fish was linked to more [3].

Stroke. A 2012 meta-analysis of 16 studies (402,127 people) found that eating fish about once a week was linked to a 14% lower stroke risk (HR 0.86), with the link stronger for ischaemic stroke [9]. Cohort data also linked 2 to 4 servings a week with a lower cerebrovascular risk (RR 0.94) [5].

Dementia and memory. A 2024 meta-analysis of 35 studies found that the highest fish eaters had an 18% lower risk of dementia (RR 0.82) and a 20% lower risk of Alzheimer's disease (RR 0.80) [10]. A 2016 meta-analysis of 21 cohorts estimated that each extra serving a week was linked to a 5% lower dementia risk (RR 0.95) [11].

Triglycerides. Randomised trials of lean fish found that it lowered triglycerides compared with no fish, though the trials were small and did not measure heart events [12]. A network analysis of 66 trials ranked fish first for lowering triglycerides, but rated the evidence as low quality [13].

Death from any cause. Fatty fish was linked to a 3% lower risk of death from any cause (RR 0.97) [1]. In a review of cohorts of people with type 2 diabetes, the highest fish eaters had a 14% lower risk of death (RR 0.86) [14]; this link was not significant in studies that also adjusted for total energy intake.

Mood and depression. A 2025 meta-analysis of 35 observational studies linked fish eating to a 21% lower depression risk (RR 0.79), with the link significant only at about 68 g a day or more [15]. A 2025 review of diet and depression found no effect for most food groups, including fish, and rated the overall evidence as very low [16].

Brain development in pregnancy. In a Seychelles cohort of 779 mother-child pairs, prenatal mercury from ocean fish was not linked to developmental harm at age 9, the authors concluded. Only two of many measures were linked to mercury: one worse in boys, one better [17]. In a French cohort of 6,780 to 11,278 children, higher fish intake in pregnancy was linked to slightly higher developmental scores at 3.5 years [18]. A Norwegian cohort linked maternal seafood intake to better language scores at age 5 [19].

Doses used in studies

Studies compare intake categories rather than set a dose. Examples: about 1 serving a week in a dementia cohort review [11]; 2 to 4 servings and about 5 servings a week for stroke [9]; 100 g a week for acute coronary syndrome [8]; 20 g a day for CHD [6]; 150 g a day for cognitive outcomes [10]; 2 to 3 servings a week as the lowest-risk level in people with type 2 diabetes [14]. In pregnancy, more than 3 times a week was studied in a 26,184-woman cohort [20]. In one Seychelles cohort, mothers ate about 8 fish meals a week, averaging 106.8 g a day [21]. These are study exposures, not recommendations.

Safety and interactions

Mercury and other contaminants. Fish concentrations differ by species: swordfish had the highest mercury level in one Spanish sample [22]. In that study, average total mercury intake for the population was below the tolerable weekly limit, but the estimated methylmercury intake for boys was above it [22]. That study estimated exposure only and did not measure health effects.

Pregnancy and young children. A pooled analysis of 26,184 women found that eating fish more than 3 times a week in pregnancy was linked to faster infant weight gain (aOR 1.22) and to overweight or obesity at age 6 (aOR 1.22); the authors said their findings matched the US FDA and EPA fish limit [20]. A 2016 review of the FDA assessment estimated that two seafood meals a week in pregnancy could add about 3.3 IQ points by age 9. This was a modelled estimate, not a trial result [2]. In the Seychelles cohorts, most child tests showed no link with prenatal mercury [17], and a later cohort found maternal fish intake was neither beneficially nor adversely associated with child neurodevelopment [21]. Eating fish in pregnancy was not linked to child asthma in a pooled cohort of 60,774 mother-child pairs [23]. No data were found on breastfeeding.

Fried fish. Fried fish was probably linked to higher risk of cardiovascular events and heart attack [3].

Side effects, upper limits and medicines. The studies read for this page did not measure side effects of eating fish, and no upper intake limit for whole fish was found. No study of whole-fish intake and medicine interactions was found. Fish-oil supplements are covered on their own topic page.

Kidney, liver and bleeding disorders. No data were found for people with these conditions in the studies read.

Caveats and open questions

Most heart, stroke, dementia and depression findings come from observational studies, which cannot show that fish causes the difference. An umbrella review of these meta-analyses noted possible confounding by other differences between people [24]. That review rated the evidence for Alzheimer's disease and for stroke as limited [24].

Some results conflict. A 2021 review found no significant link between fish and stroke [3]. Eating fish showed no clear effect on blood pressure or cholesterol: lean-fish trials found no significant change in total cholesterol or lipoproteins, and fish intake was not linked to new high blood pressure in cohort studies [12][25][26]. The depression evidence points in different directions [15][16].

Trials of fish-oil supplements, not whole fish, found no effect on cerebrovascular disease [5]. Three large US cohorts found no lower suicide risk with fish or omega-3 intake [27]. We found no trials of whole-fish intake that measured heart attacks, stroke or dementia. No relevant ongoing trials were found on ClinicalTrials.gov. One 2024 dementia meta-analysis includes an author employed by a food company [10].