Rats were fed purified diets containing 10% fat with constant (n-6):(n-3) polyunsaturated fatty acids [(n-6):(n-3); 2.3-2.6] and polyunsaturated:saturated fatty acids (1) ratios. This was obtained with alpha-linolenic acid, eicosapentaenoic acid and docosahexaenoic acid added at 1 g/100 g diet. Eicosapentaenoic acid and docosahexaenoic acid were added as the ethyl esters. The concentration of plasma cholesterol in rats fed docosahexaenoic acid was significantly lower than in those fed alpha-linolenic acid. The concentration of plasma triglyceride was significantly lower in rats fed eicosapentaenoic acid than in those fed docosahexaenoic acid. Docosahexaenoic acid significantly reduced hepatic cholesterol compared with alpha-linolenic acid and eicosapentaenoic acid. Both eicosapentaenoic acid and docosahexaenoic acid decreased hepatic triglyceride compared with alpha-linolenic acid, but this effect was more pronounced in the docosahexaenoic acid group. There was no significant difference in fecal excretion of neutral and acidic steroids and apparent fat absorption. In rats fed docosahexaenoic acid, the proportion of arachidonic acid in liver microsomal phosphatidylcholine was lower than in those fed eicosapentaenoic acid. The same tendency was observed in plasma, platelet and aortic phosphatidylcholine and liver microsomal phosphatidylethanolamine and phosphatidylinositol. Dietary docosahexaenoic acid, but not eicosapentaenoic acid, significantly decreased aortic production of prostacyclin compared to alpha-linolenic acid, whereas platelet aggregation by collagen was not affected by the difference in dietary (n-3) polyunsaturated fatty acids.