We examined the cross-sectional relationships between the frequency of habitual fish consumption, plasma phospholipid fatty acids, and serum lipids and lipoproteins in 152 men and women. There was a significant association between fish consumption starting from 1 dish/wk and plasma n-3, n-6, and n-9 fatty acids. Plasma eicosapentaenoic acid (EPA; 20: 5n-3) reflected fish consumption to a greater extent than did docosahexaenoic acid (DHA;22:6n-3). Triglycerides decreased (P less than 0.05) with fish consumption. In multivariate analysis in which anthropometric and lifestyle factors were controlled for, EPA correlated inversely with triglycerides (P less than 0.05) and positively with high-density-lipoprotein (HDL) cholesterol and apolipoprotein A-I (both P less than 0.001). In contrast, DHA did not correlate with triglycerides and showed negative associations to HDL cholesterol and apolipoprotein A-I (both P less than 0.001). Platelet phospholipid EPA, but not DHA, was associated with lower triglyceride and higher HDL-cholesterol concentrations (both P less than 0.05). This study suggests that long-term intake of small amounts of fish has biological effects, and that EPA and DHA have divergent relations with lipoprotein metabolism.