Childhood obesity is one of the major public health challenges of the 21st century. Randomized interventions have shown promising evidence of long-term effects from nutrition services on children's health. However, it is yet unclear whether scaled-up programs, such as school meals, prevent children's obesity risk. I implement a fuzzy regression discontinuity approach on national administrative data to estimate the short- and medium-run effects of the Chilean school meal program (SMP) on children's body mass index (BMI) z-scores. Girls' eligibility in first grade reduces local average obesity prevalence by 4 percentage points. Continued eligibility reduces boys' local average obesity prevalence in fifth grade by .10 percentage points. Effects concentrate among children with high BMI z-scores and are partly driven by improvements in the nutritional quality of meals. Children attending schools providing psychosocial support exhibit larger benefits from SMP eligibility, consistent with spillover effects from the integration of stimulation and nutrition interventions.