BackgroundThis study investigated whether self-reported race/skin colour and perceived racial discrimination predict higher obesity incidence after approximately 4-year follow-up of the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil). We also investigated whether these associations are modified by educational level.MethodsFollowing exclusion of individuals defined as obese (body mass index ≥30 kg/m²) at baseline, associations between race/skin colour and obesity incidence between the first (2008–2010) and second (2012–2014) visits were investigated in 10 130 participants. Next, associations between perceived racial discrimination and obesity incidence among black (n=1532) and brown (n=2958) individuals were investigated separately. Racial discrimination (yes/no) was assessed using the Lifetime Major Event Scale. Logistic regression models adjusted for age, sex and research site were used. All analyses were stratified for educational level.ResultsObesity risk was higher in Blacks with high education compared with white individuals to the same education level (OR: 2.22; 95% CI 1.62 to 3.04) following adjustments. After adjustments, obesity incidence was higher among black individuals reporting racial discrimination compared with peers who did not report this experience, but only among the low education group (OR: 1.64; 95% CI 1.08 to 2.51). No statistical association with perceived discrimination was observed among brown individuals.ConclusionResults are congruent with findings from other studies reporting associations between racial inequality and obesity incidence and also suggest racial discrimination may be one of the mechanisms leading to such inequalities. Also, it supports the paradox theory by which education modify the association in distinct directions.