OBJECTIVE Nutritional epidemiology is increasingly shifting its focus from studying single nutrients to the exploration of the whole diet utilizing dietary pattern analysis. We analyzed associations between habitual diet (including macronutrients, dietary patterns, biomarker of fish intake) and lipoprotein particle subclass profile in young adults. RESULTS Complete dietary data (food-frequency questionnaire) and lipoprotein subclass profile (via nuclear magnetic resonance spectroscopy) were available for 663 subjects from the population-based FinnTwin12 study (57% women, age: 21-25 y). The serum docosahexaenoic to total fatty acid ratio was used as a biomarker of habitual fish consumption. Factor analysis identified 5 dietary patterns: "Fruit and vegetables", "Meat", "Sweets and desserts", "Junk food" and "Fish". After adjustment for sex, age, body mass index, waist circumference, physical activity, smoking status and alcohol intake, the "Junk food" pattern was positively related to serum triglycerides (r = 0.12, P = 0.002), a shift in the subclass distribution of VLDL toward larger particles (r = 0.12 for VLDL size, P < 0.001) and LDL toward smaller particles (r = -0.15 for LDL size, P < 0.001). In addition, higher scores on this pattern were positively correlated with concentrations of small, dense HDL (r = 0.16, P < 0.001). Habitual fish intake associated negatively with VLDL particle diameter ("Fish" pattern and biomarker) and positively with HDL particle diameter (biomarker). CONCLUSIONS Our results suggest that in young adults, higher habitual fish consumption is related to favorable subclass distributions of VLDL and HDL, while junk food intake is associated with unfavorable alterations in the distribution of all lipoprotein subclasses independent of adiposity and other lifestyle factors.