BACKGROUND Since children are generally not achieving desirable levels of physical activity (PA), schools can provide an opportunity to increase their active time. This study aimed to describe individual‐ and social‐level variables associated with PA at school recess in a culturally diverse primary school. METHODS This observational cross‐sectional study included 131 children attending primary school. PA data were measured via accelerometry during school recess and analyzed in relation to data on sociodemographics, anthropometrics, and PA outside school. Data were analyzed using generalized linear modeling and logistic regression. FINDINGS Girls were more sedentary and engaged in less moderate to vigorous physical activity (MVPA) than boys (p < .001), while older girls were less active than younger girls (p < .001). Also, non‐Caucasian children were less sedentary (p < .001) and engaged in more MVPA (p = .007). Belonging to ethnic minorities was also associated with lower socioeconomic status, higher body mass index, and less PA outside of school and on weekends. CONCLUSIONS Girls (especially older girls) and children from lower socioeconomic contexts should be prioritized when promoting PA. IMPLICATIONS FOR SCHOOL HEALTH POLICY School recess can be used as an effective strategy aimed at increasing PA and reducing social inequalities.