Although a large literature examines the effect of non-parental child care on preschool-aged children's cognitive development, few studies deal convincingly with the potential endogeneity of child care choices. Using a panel of infants and toddlers from the Birth cohort of the Early Childhood Longitudinal Study (ECLS-B), this paper attempts to provide causal estimates by leveraging heretofore unrecognized seasonal variation in child care participation. Child assessments in the ECLS-B were conducted on a rolling basis throughout the year, and I use the participation "dip" among those assessed during the summer as the basis for an instrumental variable. The summer participation "dip" is likely to be exogenous because ECLS-B administrators strictly controlled the mechanism by which children were assigned to assessment dates. The OLS results show that children utilizing non-parental arrangements score higher on tests of mental ability, a finding that holds after accounting for individual fixed effects. However, the instrumental variables estimates point to sizeable negative effects of non-parental care. The adverse effects are driven by participation in formal settings, and, contrary to previous research, I find that disadvantaged children do not benefit from exposure to non-parental child care settings.