In two long-term feeding trials in seawater, Atlantic salmon were fed EPA+DHA in graded levels, from 1·3 to 7·4 % of fatty acids (FA, 4-24 g/kg feed) combined with approximately 10 % 18 : 3n-3, at 6 and 12°C. Dietary EPA appeared to be sufficient in all diet groups, as no differences were seen in polar lipid tissue concentrations of either the brain, retina or erythrocytes. For DHA, a reduction in tissue levels was observed with low dietary supply. Effects on brain DHA at ≤1·4 % EPA+DHA of dietary FA and retina DHA at ≤2·7 % EPA+DHA of dietary FA were only observed in fish reared at 6°C, suggesting an effect of temperature, whereas tissue levels of n-6 FA increased as a response to increased dietary n-6 FA in both the brain and the retina at both temperatures. DHA levels in erythrocytes were affected by ≤2·7 % EPA+DHA at both temperatures. Therefore, DHA appears to be the limiting n-3 FA in diets where EPA and DHA are present in the ratios found in fishmeal and fish oil. To assess the physiological significance of FA differences in erythrocytes, the osmotic resistance was tested, but it did not vary between dietary groups. In conclusion, ≤2·7 % EPA+DHA of FA (≤9 g/kg feed) is not sufficient to maintain tissue DHA status in important tissues of Atlantic salmon throughout the seawater production cycle despite the presence of dietary 18 : 3n-3, and effects may be more severe at low water temperatures.