The effect of fish oil rich in eicosapentaenoic (EPA) and docosahexaenoic (DHA) acids on serum lipoprotein concentrations is not clear, and it is not known whether EPA and DHA are similarly related to serum lipid or lipoprotein levels. We conducted a randomized, 10-week, dietary supplementation trial in which the effects of 6 g per day of 85% EPA and DHA were compared with 6 g per day of corn oil in 156 men and women. Multivariate analyses were used to assess independent relations between plasma phospholipid EPA and DHA and serum lipoprotein levels. In the fish oil group triglycerides fell 21% (p less than 0.001) and high density lipoprotein cholesterol (HDL-C) rose 3.8% (p less than 0.05). In the corn oil group triglycerides did not change, but HDL-C rose 6.1% (p less than 0.01). Compared with fish oil, apolipoprotein A-I (apo A-I) rose 5.1% after corn oil intake (p less than 0.05). Plasma EPA and DHA levels rose after fish oil intake and fell after corn oil intake (all p less than 0.001). The change (delta) in EPA was inversely correlated with delta triglycerides (p = 0.035) and positively correlated with delta HDL-C and delta apo A-I (both p less than 0.001) in the multivariate analyses. In contrast, delta DHA was not correlated with delta triglycerides but was inversely correlated with delta HDL-C and delta apo A-I (both p less than 0.001). Standardizing for DHA removed the difference in apo A-I levels between groups. This study suggests that EPA and DHA are divergently associated with HDL, possibly through different mechanisms.