The Every Student Succeeds Act's evidence requirements mandate the use of research in the selection of school improvement interventions, with the aim of ensuring that schools and districts invest their efforts and funding more wisely. This study of eight states presents three different approaches to the evidence requirements: using lists of pre-sanctioned evidence-based interventions, training schools and districts to evaluate the research supporting potential interventions, and building local evidence of effectiveness. Through interviews with state administrators, I show how each approach relied on different understandings and prioritizations of research rigor and local relevance.