In this study, the relationships between dietary fat [as measured by serum cholesterol ester fatty acids (CE-FA)], age, smoking, body mass index, and serum lipids were analyzed in 151 subsistence horticulturalists, aged 20-86 yr, from Kitava, Trobriand Islands, Papua New Guinea. Their diet consists of tubers, fruit, coconut, fish, and vegetables with a negligible influence of western food and alcohol. Total fat intake is low [21% of energy (en%)], while saturated fat intake from coconuts is high (17 en%, mainly lauric and myristic acid). In multivariate analysis, 11-43% of the variation of the serum lipoprotein composition was explained by CE-FA, age, and smoking habits. The proportion of CE20:5n-3 explained much of the variation of triglycerides (TG, negative relation) and high density lipoprotein-cholesterol (HDL-C, positive) in both sexes and serum apolipoprotein A1 (ApoA1, positive) in the males. CE16:0 was positively related to TG and negatively related to HDL-C and ApoA1 in both sexes, and in males it related negatively to total cholesterol (TC) and low density lipoprotein-cholesterol (LDL-C). In males, negative relationships were present between CE18:2n-6 and TC and between CE14:0 and serum lipoprotein(a). Smoking was independently associated with lower ApoA1 in both sexes and with lower HDL-C and higher TG, TC, LDL-C, and apolipoprotein B in males. In conclusion, marine n-3 fatty acids and linoleic acid showed the same potentially beneficial relationships with lipoproteins and apolipoproteins as in western populations. The relations of palmitic acid to serum lipids may be explained in terms of endogenous fat synthesis at a low-fat intake, rather than reflecting its relative intake.