Omega-3 Fatty Acids
(including Fish Oil)
(updated 2/2013)

Bottom line: the research is clear. This stuff works. Not very powerfully, and it takes a while to get in there. So if you're going to do it, give it two months to evaluate. 

If you're going to go this route, buy one large bottle of pills or liquid  (wait, wait, it's lemon flavored, you can do this) and finish it; you'll need to go at least that long to see if you're clearly better. 

Omega-3 fatty acids, given in the form of fish oil tablets, were shown in a 1999 controlled trialStoll to maintain mood stability far better than an olive oil placebo. This was a very exciting result, especially as the trial was inspired by "bench chemistry" --  omega-3 fatty acids were recognized to have intracellular effects similar to lithium and valproate.  

There are two "omega-3" fatty acids in fish oil: EPA (eicosapentaenoic) and DHA (docusahexaenoic).  Some studies in people with depression or bipolar disorder have used EPA alone, including one of the most recent with the strongest positive results; but one of the larger earlier studies using EPA alone showed no benefit.  Many of the other studies shown below, including the first one from Harvard, used fish oil tablets which provided both EPA and DHA. For now it seems safe to conclude that having both does not interfere with the antidepressant and mood stabilizing actions; and that EPA alone might be sufficient. 

You may have heard of omega-3's or fish oil as having some benefit in reducing heart disease as well.  That research is a little farther along than the mood research. I have not followed it closely but it seems that the focus there is more on DHA than EPA. This means that if you want both benefits, you may need to stick with a source that has both EPA and DHA. Luckily, it turns out that the cheapest version has both, and has been lowest in the tests of mercury (undetectable levels). So for now it looks like the cheapest source may be the best -- an unusual situation in my business! Details on which kind to consider, and how many pills you'll need, can be found on the "more fish (oil) stories" page.  For another excellent source of information, a bit more technical but more complete, see the Mayo Clinic pages on omega-3 fatty acids. 

Do they work? 

[Update 1/2004: a statistical study also recently showed fish-eating countries have lower bipolar rates than those where fish is not common in the diet.Noaghiul  Another study looked at whether patients with bipolar disorder have lower omega-3 levels than "healthy controls", and found lower DHA but not EPA levels in their blood cell walls.  Even with a small sample they found a pretty big difference.Chiu]

Here's a summary of fish oil research regarding use in mood disorders, updated 10/2011 (some studies of weak statistical quality are not included); followed by some details and conclusions.  Note that in all the bipolar studies, and several of the depression studies, the omega-3 fatty acid dose was added to prior medications.  These data should not be interpreted to indicate that O-3's alone are a treatment for mood disorders.  

Author/Link Condition Studied total # of patients; placebo control? EPA/DHA/fish oil (grams) Outcome
Stoll, 1999 Bipolar, recently in remission 80/placebo 9.6 grams of fish oil  Relapse rate sharply reduced in O-3 group
Keck, 2006 Bipolar Depression 120/placebo 6 grams EPA No difference from placebo
Frangou, 2006 Bipolar  Depression
(rather mild)
75 divided into three groups, incl. placebo; added to ongoing medications 1 gram EPA versus
2 grams EPA
EPA significantly better than placebo; 1 gram no better than 2; no increase in mania scores on EPA
Sagduyu, 2005 Bipolar 37/ no placebo 0.5 to 6 grams; mostly 1-2 grams  less irritability
Osher, 2005 Bipolar Depression 12/no placebo 1-2 grams EPA 8 of 10 responded; no mania increase
Zanarini, 2003 "Borderline" 30/placebo 1 gram EPA Decreased aggression, depression (if use particular statistical tweaking)
Jazaveri, 2008 Depressed 60/placebo 1 gram EPA or fluoxetine (Prozac) or both EPA equivalent to fluoxetine, combination better than either alone
Nemets, 2002 Depressed 20/placebo 2 grams EPA Highly significant benefits by week 3
Peet, 2002 Depressed 70/placebo 1, 2, and 4 grams EPA only 1-gram group did best; 2 gms did not improve, 4 gms trended better
Marangell, 2003 Depressed 35/placebo 2 grams DHA At 6 weeks, a trend but not significant benefit
Silver, 2005 Depressed 77/placebo 8 grams of fish oil No benefit; used "in addition to usual therapy", however
Su, 2003 Depressed 28/placebo (estimated)4.4 grams EPA/ 2.2 grams DHA Significant benefits versus placebo by week 4, bigger yet by week 8
Chiu, 2003 Depressed, pregnant case report 4 grams EPA/ 2 grams DHA Improvement by week 4, better yet by week 6
Su, 2008 Depressed, pregnant 36/placebo  2.2 grams of EPA and 1.2 grams of DHA 62% responded, versus 27% on placebo; and remission, 38% vs. 18% on placebo. 
Puri, 2001 Depressed, 7 yrs of rx not working Case report 4 grams of fish oil "structural brain changes" reported (smaller ventricles, which theoretically is consistent with brain growth)
Amminger et al.  Schizophrenia prevention 41 "subthreshold psychosis"/ 40 placebo 1.2 grams/day of omega-3's total 2 out of 41 taking fish oil developed psychosis; vs. 11 of 40 in the control group

A more detailed look, study by study, is available on my Fish Oil Research page. Also see there the big review by Sublette, 2011, regarding dose

What dose should I take?

Update 2011: a big review of all these studies says 1 gram of EPA per day may be the best standard dose to shoot for. As much as 2 grams of EPA is okay but not clearly better.  More than 2 grams  might actually produce less benefit (not to mention how many pills that would require, and their cost). 

 How many pills do you need to get 1 gram of EPA?  That depends on how concentrated the pill is, and how much EPA per pill. Cheaper means less concentrated, which means more pills per day to reach this target dose.  Pay more, you might be able to take fewer pills.   

If you're up to it (this is algebra! but simple) the review also suggests that the ratio of EPA to DHA (the two omega-3 fatty acids) really matters. Their analysis shows that of these two omega-3's in each pill, at least 60%  must be EPA. Less than that and it does not work better than a placebo; in fact in several studies, less than 60% EPA was worse than a placebo. 

Read the label very carefully and use a calculator to make sure that EPA is at least 60% of the total omega-3 per pill.  Tricky.  See my fish stories page for the oily details on the cheapest way to do this


There are basically no known risks with fish oil. [Update August 2013: there are new concerns about prostate cancer risk. Men should see my separate page about omega-3's and prostate cancer.] Three which have been discussed can be nearly dismissed, as shown here. This does not mean there are no risks; it means there are no significant known risks.  Notice the difference. On the other hand, one cannot say anything as nearly reassuring about any of our regular medications. Here are three risks which have been discussed. 

  1. Mercury and other heavy metals
  2. Hypomania
  3. Increased risk of easy bleeding if taking other medications known to do likewise. 

1. Mercury. One of the cheapest fish oils, shown in the following table, has "undetectable levels" so you could stop worrying about it right there.  But being reassured like that doesn't work for a lot of people, so here are more details if you like. Otherwise, skip forward to #2)

Some would not entirely trust the United States Food and Drug Administration (FDA) for advice, but this at least gives you a yardstick of some kind. The fetal brain is more susceptible to mercury-induced damage, so here here is what the FDA tells pregnant mothers (not everyone, you see, just pregnant mothers):  

The table below gives mercury in a different unit, making comparison with the above yardstick difficult.  A review article on fish oil for the American Heart Journal in March of 2006DeFillipis summarized this issue by noting that the FDA believes 10 mcg/L, which is found in some fish oils (see table below) is "negligible". Less than 6 mcg/L, as with the Kirkland brand, means "undetectable" using these laboratory tests.   

Brand Name Mercury level 
CVS 10
Kirkland <6
Nordic Ultimate <6
Omega-Brite 12
Sundown <6

Consumers Reports (CR) also has conducted tests. Their conclusion: "None showed evidence of spoilage, and none contained a significant amount of mercury, PCBs, or dioxin.  CR found no significant differences in the quality or purity of these supplements. Therefore, consumers can choose them based on price." (quoting from a summary in a senior newsletter so you don't have to pay for the Consumer Reports report itself, which of course you could pursue if interested).  

Update 6/2008: you can also view a summary of risks from the Environmental Defense Fund, which looked at PCB's and environmental factors as well as mercury.  Costco's Kirkland brand comes out fine again. A colleague points out that the doses of fish oil studied in many of these lab tests are much lower than those which are now commonly recommended for the treatment of depression.  Therefore, even low levels of mercury or other heavy metals might still be relevant, if these pills were taken at high doses for a long time. Two recommendations emerge from this concern:

  1. Since at least in theory there may still be some risk from heavy metal exposure if you're taking high doses, you ought to be quite certain that taking it really make the difference, that it really helps you. 
  2. If you are going to use it for a long time, you should probably use a version of fish oil which has the lowest levels of heavy metals you can find.

2. Hypomania.  Notice the emphasis in the research studies on depression, versus bipolar disorder.  There is a report of hypomania developing while on omega-3'sKinrys (and in the Harvard study as wellStoll), although whether the fatty acids were the "cause" of this episode is of course not certain. I'm watching for more such reports (update 2011: no more reports, no patients of mine ever showing such signs I could associate with the fish oil...). 

3. Bleeding risks. (updated 11/09) There are reports of fish oil interfering with blood coagulation if you taking aspirin or coumadin (warfarin).Buckley  But a reader who is not on coumadin wrote with her experience of increased bleeding (vaginal, nose bleeds) on fish oil. She discovered thereby that she has a mild form of a fairly common bleeding disorder (von Willebrand's). It just didn't show up until she started the fish oil. Thanks for the heads up, Ms. M. 

Side Effects

See Risks above first, there are some; but as for side effects, there is really only one: fish burps ("When I burp I feel like a seal").  This is easily managed by taking the pills with meals. If that isn't sufficient to stop the problem, put the pills in the refrigerator, or freezer, between meals. This makes them sort of an "extended release" capsule. Watch out for condensation inside the bottle if you live in a moist climate. You could just keep a week's supply or so in a separate bottle that you open every day. 

Conclusions (rewritten as of 10/2011):  

Do I recommend them to my patients? Yes, rather commonly, especially when we're looking for all possible anti-depression tools to use, rather than have to use a typical antidepressant.  I still have yet to see a clear-cut stunning improvement like those case reports in the table above, but because the risk appears to be so minimal, in many cases it seems like it's worth a try.  

 For small details including , pill type, flaxseed oil and any other tidbits I accumulate, here are more of my fish (oil) stories.