Under the "No Child Left Behind Act" (NCLB), in theory, schools on the margin for meeting AYP face strong short-term incentives to increase students' pass rates on specific exams, and may change their behavior accordingly. Using a comprehensive, national, school-level data set concerning schools' AYP status, student characteristics, and student test score performance, as well as NCLB test score performance and school characteristics, the authors predict which schools were near the margin for failing to meet their state's AYP standards in math and reading. Variance in state policies creates several cases where schools near the margin for satisfying their "own" state's AYP requirements would have almost certainly failed or almost certainly passed AYP if they were located in other states. Using the nationally representative Early Childhood Longitudinal Survey (ECLS), the authors examine how NCLB incentives affect students' academic achievement and non-academic outcomes, and school resource allocation. States vary widely in the percent of schools that fail or struggle to meet AYP, with cross-state variation in student academic aptitude or in exam difficulty explaining relatively little of this variation. Rather, cross-state variation in AYP failure rates is due to states' choice of policy parameters concerning AYP rules, such as the minimum student enrollment size required for student subgroups' pass rates to contribute to a school's AYP status, the size of confidence intervals and the "safe harbor" rules that effectively lower the minimum pass rates required for smaller subgroups to pass AYP. Our very preliminary results suggest that NCLB pressure influences student and staff attitudes and teachers' time use and instructional strategies but has little net effect on mean student test score growth on low-stakes exams. Data Sources for First Stage Analyses are appended. (Contains 5 tables, 4 figures and 21 footnotes.) [This paper was researched with assistance from Ben Lockwood, Julia Zhou, and Molly Alter. Additional funding for this research was provided by the Columbia University Institute for Social and Economic Research and Policy.]