Interventions targeting early childhood hold promise for reducing the intergenerational transmission of poverty. Results from a randomized evaluation of a preschool construction program in Cambodia suggest caution. Overall impacts on early childhood outcomes are small and insignificant. Impacts on cognition are negative for the cohort with highest program exposure, with the largest negative effects among children of poorer and less educated parents. The results are explained by substitution from primary to preschool and differences in demand responses to preschools between more and less educated parents. Context, program specifics, and behavioral responses can hence lead to perverse effects of well-intentioned interventions. (author abstract)