ObjectiveFew large-scale studies have provided population-based estimates of hair cortisol levels and its determinants. Hair cortisol and potential determinants were measured in children and their mothers in a population-based sample in a Brazilian city with large variations in socioeconomic conditions. MethodsWe used data from the 4-year follow-up of the 2015 Pelotas (Brazil) Birth Cohort Study. Hair samples were collected by trained fieldworkers to analyze average levels of cortisol over a 3-month period. Four groups of variables were tested as potential determinants: hair characteristics (natural color, treatment, type, and frequency of wash), use of corticosteroids and oral contraceptives, sociodemographic factors (sex, age, skin color, socioeconomic level, maternal relationship, pregnancy, daycare enrollment), maternal perceived stress, and substance exposure (smoking and illicit drug use). Linear regression with log transformation was used to test associations. Results3235 children and 3102 mothers were analyzed (80.7% and 77.4% of those interviewed when children were 4 years of age, respectively), for whom sufficient hair was collected for cortisol analysis. The median of hair cortisol concentration was 7.8 pg/mg (IQR = 5.6 – 11.0) for children, and 5.6 pg/mg (IQR = 4.2 – 7.8) for mothers. In adjusted models, sex and socioeconomic level were associated with child cortisol levels. For mothers, hair cortisol levels were associated with socioeconomic level, skin color, age, hair treatment and hair natural color. ConclusionThis study provides estimates of hair cortisol levels in a diverse population in a upper-middle income country. Although just a few predictors were associated with maternal/child cortisol levels, socioeconomic level was the key variable that should be incorporated in studies using hair cortisol to measure biological manifestations of stress, but other variables, such as some hair and sociodemographic characteristics are important to consider when using hair cortisol.