Objective: To assess the association between breast feeding and blood lipid levels in adolescence. Design: Population based prospective birth cohort study. Setting: City of Pelotas, Brazil. Subjects: All hospital births taking place in 1982; 79% of all males (n = 2250) were followed up for 18 years, and 2089 blood samples were available. Interventions: None. Main outcome measures: Total cholesterol and fractions (very low density lipoprotein cholesterol (VLDL), low density lipoprotein cholesterol (LDL), high density lipoprotein cholesterol (HDL)), LDL/HDL ratio, serum triglycerides. Results: Three breast feeding variables were studied: total duration of breast feeding, duration of exclusive or predominant breast feeding, and ever compared with never breast fed. Adjusted analyses were controlled for family income, household assets index, maternal education, maternal pre-pregnancy body mass index (BMI), skin colour, birth weight, gestational age, maternal smoking during pregnancy, and adolescent BMI, and behavioural variables (fat content of diet, physical activity, smoking, and alcohol drinking). Only one association reached borderline significance (p = 0.05): LDL cholesterol was slightly higher among never (mean 41.0 mg/dl; 95% CI 39.4 to 42.7) than among ever breast fed men (38.6 mg/dl; 95% CI 38.6 to 40.3), in the adjusted analyses. All other associations were not significant (p>0.09). There was no evidence of effect modification according to preterm status, intrauterine growth retardation, socioeconomic level, growth velocity in the first two years of life, or nutritional status at 2 years of age. Conclusions: There was no clear association between breast feeding duration and serum lipid concentrations at the age of 18 years in this sample of Brazilian men.T he availability of results from long term birth cohort studies has permitted the assessment of the long term health consequences of early life exposures. A recent publication by British authors reported on two parallel randomised controlled trials in which preterm newborns were allocated to receive either donated banked human milk or formula (study 1) or two different types of formula (study 2). At the ages of 13-16 years, adolescents who had been randomised to banked breast milk had lower low density lipoprotein (LDL) to high density lipoprotein (HDL) cholesterol ratio (mean difference 0.34 (14% lower), 95% CI 20.67 to 20.01; p = 0.04) than those given preterm formula. There was also a dose-response relation with the intake of breast milk in infancy, thus suggesting a long term impact of breast milk feeding on the risk of atherosclerosis.