BACKGROUND: Most preschool children in Western industrialized countries attend child care during the day while parents work. Studies suggest that child care may be stressful to young children, perhaps because they still lack the social skills to interact daily in a group setting away from parents. This gap in social abilities may be greater for children in lower-income families, who may face more adversity at home, with fewer resources and more social isolation. METHODS: We conducted a cluster-randomized controlled trial in 2013-2014 to test whether a social skills intervention led by early childhood educators within the child care center could reduce diurnal cortisol levels to more typical patterns expected of children this age. We randomized 19 public child care centers (n = 361 children) in low-income neighborhoods of Montreal, Canada, to either: 1) the Minipally program - intervention group (n = 10 centers; 186 children), or 2) waiting list - control group (n = 9 centers; 175 children). Saliva samples for cortisol levels were collected 3 times/day, pre- and post-implementation. The Minipally puppet program consists of 2 workshops/month for 8 months for the development of social skills and self-regulation in 2-5-year-olds, with reinforcement activities between workshops. Educators received 2-days' training and 12 h' supervision in Minipally. RESULTS: Linear mixed models for repeated measures revealed a significant interaction between intervention status and time of day of cortisol sampling (beta = -0.18, p = 0.04). The intervention group showed patterns of decreasing diurnal cortisol secretion (beta = -0.32, p < 0.01), whereas the control group showed increasing slopes (beta = 0.20, p < 0.01). Moreover, family income was a moderator; children in lower-income families benefited most from the intervention. CONCLUSION: Results suggest that a social skills training program, when integrated into a preschool education curriculum, can foster an environment more conducive to typical childhood patterns of cortisol secretion.