Much of child care research has focused on the effects of the quality of care in early childhood settings on children's school readiness skills. Although researchers increased the statistical rigor of their approaches over the past 15 years, researchers' ability to draw causal inferences has been limited because the studies are based on nonexperimental designs. The purpose of the present study was to demonstrate how an instrumental variables approach can be used to estimate causal impacts of preschool center care quality on children's academic achievement when applied to a study in which preschool curricula were randomly assigned across multiple sites. We used data from the Preschool Curriculum Evaluation Research Initiative (PCER; n = 2,700), in which classrooms or preschools were randomly assigned to that grantee's treatment curriculum or "business as usual" conditions in 18 research sites. Using this method, we demonstrate how developmental researchers can exploit the random-assignment designs of multisite studies to investigate characteristics of programs, such as preschool center care quality, that cannot be randomly assigned and their impacts on children's development. We found that the quality of preschool care received by children has significant, albeit modest, effects on children's academic school readiness, with effect sizes of .03 to .14 standard deviation increases in academic achievement associated with a 1 standard deviation increase in quality. Applications and potential policy implications of this method are discussed. (author abstract)