Weanling rats were fed on high-fat (178 g/kg) diets which contained 4.4 g alpha-linolenic (ALA), gamma-linolenic, arachidonic (ARA), eicosapentaenoic (EPA), or docosahexaenoic acid (DHA)/100 g total fatty acids. The proportions of all other fatty acids, apart from linoleic acid, and the proportion of total polyunsaturated fatty acids (PUFA) (approximately 35 g/100 g total fatty acids) were constant, and the n-6 to n-3 PUFA ratio was maintained as close to 7 as possible. The fatty acid compositions of the serum and of spleen leukocytes were markedly influenced by that of the diet. Prostaglandin E-2 production was enhanced from leukocytes from rats fed the ARA-rich diet and was decreased from leukocytes from the EPA-or DHA-fed rats. Replacing dietary ALA with EPA resulted in diminished ex vivo lymphocyte proliferation and natural killer (NK) cell activity and a reduced cell-mediated immune response in vivo. In contrast, replacing ALA with DHA reduced ex vivo lymphocyte proliferation but did not affect ex vivo NK cell activity or the cell-mediated immune response in vivo. Replacement of a proportion of linoleic acid with either gamma-linolenic acid or ARA did not af feet lymphocyte proliferation, NK cell activity, or the cell-mediated immune response. Thus, this study shows that different n-3 PUFA exert different immunomodulatory actions, that EPA exerts more widespread and/or stronger immunomodulatory effects than DHA, that a low level of EPA is sufficient to influence the immune response, and that the immunomodulatory effects of fish oil may be mainly due to EPA.