A group of 33 mildly hypercholesterolemic men were stratified into three groups on diets closely matched except for the polyunsaturated fatty acid supplement. The first group received 14 g/day of linoleic acid (safflower oil); the second group, 9 g of alpha-linolenic acid (linseed oil); and the third group, 3.8 g of n-3 fatty acids (fish oil). Only fish oil lowered plasma triglycerides (by 24% at 6 weeks, p less than 0.05 compared to safflower oil). Very low density lipoprotein (VLDL) apoprotein (apo) B, triglyceride, and cholesterol all fell significantly with the fish-oil diet (p less than 0.01). Low density lipoprotein (LDL) cholesterol fell by 0.18 and 0.10 mmol/l, respectively, with the safflower-oil and linseed-oil diets, but rose by 0.24 mmol/l with the fish-oil diet (p less than 0.05). There was a strong correlation between the changes in VLDL triglyceride and LDL cholesterol with the fish-oil diet (r = -0.84, p less than 0.002). High density lipoprotein (HDL) cholesterol fell slightly in all three groups (p less than 0.02 with the linseed-oil diet only). However, the apo A-I/A-II ratio rose by 5% (p less than 0.05), and the HDL2/HDL3 protein ratio increased by 28% with the fish-oil diet (p less than 0.005). Fish oil reduced the capacity for transfer of cholesteryl ester between LDL and HDL by 23% (p less than 0.02 compared to baseline), reduced plasma lecithin:cholesterol acyltransferase activity by 21% (p less than 0.05), and reduced maximal stimulated thromboxane production by 9% (p less than 0.05). Thus fish oil produced three potentially beneficial changes: significant decreases in VLDL concentration and in thromboxane production and an increase in the HDL2/HDL3 ratio. The increase in the average HDL particle size probably reflected reduced cholesteryl ester acceptor capacity within the smaller pool of VLDL, as well as the decline in lipid transfer activity in plasma involving transfer protein itself, LDL, and HDL.