Do omega-3s reduce aggression? What the 29-study meta-analysis Rhonda Patrick flagged actually shows

Dr Rhonda Patrick — the biomedical scientist behind the FoundMyFitness podcast — recently called a new finding “a big deal”: across randomised controlled trials, omega-3 supplementation appears to reduce aggressive behaviour. The headline number she pointed to is striking. A meta-analysis pooling 29 studies and nearly 4,000 participants reported that omega-3s cut aggression by up to 28%, and that the benefit was broadly consistent across very different groups of people.

That is the kind of claim worth slowing down on. “Up to 28%” is a real result from a careful analysis — but it also sits on top of a modest average effect, a short measurement window, and an author who is the first to say it is not a magic bullet. Here is what the study actually did, and what it does and does not establish.

What the meta-analysis actually did

The work Patrick flagged is a 2024 meta-analysis by Adrian Raine and colleagues, published in the journal Aggression and Violent Behavior. Raine is a neurocriminologist at the University of Pennsylvania who has studied the biology of antisocial behaviour for decades, and this paper is his attempt to pool the entire randomised-trial literature on omega-3 and aggression in one place.

It draws on 29 studies (35 independent samples) run across 19 separate laboratories between 1996 and 2024, totalling 3,918 participants. Crucially, these were randomised controlled trials — people assigned to omega-3 or a placebo — rather than observational surveys, which is what lets the authors talk about effect rather than mere correlation. The trials lasted around 16 weeks on average and spanned ages from children to adults in their fifties and sixties. The analysis covered both reactive aggression (an impulsive response to provocation) and proactive aggression (the planned, instrumental kind), and found omega-3 reduced both.

How big is the effect, really?

This is where the “up to 28%” figure needs context. The analysis ran in three parts, and the standardised effect sizes came out at roughly 0.16, 0.20 and 0.28, averaging about 0.22. In the language of statistics that is a small effect — the authors themselves describe it as “modest.” The widely-quoted “up to 28%” reflects the largest of those sub-analyses, not a literal across-the-board figure, and popular coverage has at times rounded it up to 30%.

A small average effect is not the same as no effect. Raine’s argument is that for an outcome as costly as aggression and violence, even a modest, cheap, low-risk intervention is worth taking seriously. But it is honest to say the picture is “a real but small shift,” not “a supplement that makes people 28% less aggressive.” Those are different statements, and only the first is supported.

What held up across the board

The part that impressed Patrick — and that is genuinely the strongest feature of the paper — is the consistency. The beneficial effect held across the moderators the authors tested:

• Age and sex. The reduction appeared in children and in adults, and in both males and females.
• Clinical status. It showed up whether or not participants had a diagnosed condition, and across different recruitment settings.
• Dose and duration. The trials used a range of omega-3 doses and lengths, and the benefit was not confined to one narrow protocol.
• Reactive and proactive aggression alike. Both forms responded, with the clearest signals in self-reported measures.

The authors also report no evidence of publication bias, and say sensitivity analyses (re-running the numbers with different assumptions) did not overturn the result. That matters: a finding that survives those checks is harder to dismiss as an artefact of a few favourable studies.

The caveats Raine is upfront about

Raine is unusually candid about the limits of his own conclusion. A few are worth carrying with you:

• It is a short-term effect. With trials averaging about 16 weeks, the analysis speaks to what happens over a few months. It does not establish that the benefit persists for years, and the authors call for larger, longer studies.
• It is not a cure for violence. In Raine’s own words, “omega-3 is not a magic bullet that is going to completely solve the problem of violence in society.” He frames it as one modest lever among many, not a replacement for social or clinical interventions.
• Mechanism is still a hypothesis. A meta-analysis pools outcomes; it does not prove why. The leading idea — that EPA and DHA support brain structure and dampen inflammation in regions involved in impulse control — is plausible and consistent with other work, but the analysis itself does not nail it down.

His practical framing is deliberately low-key: even “an extra portion or two of fish each week could also help.” That is a long way from prescribing supplements as a behavioural treatment.

What this means in practice

For most people, this study does not change the case for omega-3 so much as add to it. The reasons to get enough EPA and DHA — cardiovascular, cognitive, anti-inflammatory — were already reasonable; a modest effect on mood-adjacent behaviour is a plus, not the headline reason to bother.

• Food first is a fair reading. Oily fish a couple of times a week supplies EPA and DHA directly. Supplements are a top-up for people who do not, not a substitute for everything else.
• Quality matters more than the claim. If you do supplement, the thing that undermines an omega-3 product is not the dose on the label but whether the oil is fresh. Long-chain omega-3s oxidise easily, and oxidised oil loses the activity you are paying for.
• Modest means realistic expectations. Treat omega-3 as one small, low-risk input — worth getting right, not worth overselling to yourself or anyone else.

Patrick’s consistent advice on choosing an omega-3 is to check the batch rather than the bottle — covered in our piece on what Dr Rhonda Patrick says about fish oil and why she checks IFOS. You can also browse IFOS-certified fish oils on Certwell, where each batch is tested for freshness, EPA/DHA content, and contaminants.