Introduction: Although epidemiologic studies have associated subclinical thyroid disorders with coronary heart disease, the role of lipoprotein levels are not well defined as a mediator or moderator of this association. Purpose: To test the association of lipoprotein particle concentration and size measured by nuclear magnetic resonance with subclinical hyperthyroidism and subclinical hypothyroidism using baseline data from the Brazilian Longitudinal Study of Health (ELSA-Brasil) of apparently healthy 5024 men and women aged 35 to 74 years old. Methods: We measured lipids and lipoproteins subfractions using a new nuclear magnetic resonance spectroscopy (NMR LipoProfile® test spectra, LabCorp, Raleigh, NC): triglyceride-rich lipoprotein particles (TRLP, containing IDL-P and VLDL-P), 3 LDL-P subclasses (large, medium and small) and 3 HDL-P subclasses (large, medium and small). Subclinical thyroid disorders were defined as TSH >4 IU/ml and TSH <0.4 IU/ml for people with normal free-thyroxine levels. We excluded from the sample participants reporting previous cardiovascular disease, those who used lipid-lowering medications and/or other drugs that interfere with thyroid function remaining 3,947 subjects. The cross-sectional relationship between subclinical thyroid diseases (independent variable) and lipids and lipoproteins (dependent variable) were evaluated by linear regression models using euthyroid subjects as the reference. Models were presented as crude and after adjustment for sociodemographic (age, sex, race, and education), cardiovascular risk factors (body-mass index, hypertension, diabetes, smoking habit, alcohol use, and physical activity) and LDL-C, HDL-C and triglycerides measured by enzymatic methods. Results: After exclusions, we detected 47 persons with subclinical hyperthyroidism (mean age 53.1 (±9.4 years; 61.7% women), 258 with subclinical hypothyroidism (mean age 52.3 (±8.9); 58.1% women) and 3,642 participants with normal thyroid function (mean age: 50.1 (±8.6); 52% women). After multivariate adjustment, subclinical hyperthyroidism was associated with a decrease in triglyceride-rich lipoprotein particles (Beta coefficient (B) -15.2; P=0.04) and with the very small triglyceride-rich lipoprotein subclass (B, -16.9; P=0.04). Subclinical hypothyroidism was associated with a decrease in small HDL-P subclass (B, 0.42; P=0.03). No association was found for any subclinical conditions with LDL-C. Conclusions: Subclinical hyperthyroidism does not show any possible association with an atherogenic profile, and subclinical hypothyroidism revealed an association with lower small HDL-cholesterol levels.