The No Child Left Behind law is flawed for many reasons, but the most important is that it is built around proficiency targets. Proficiency rates are not useful metrics of school performance because universal proficiency is not a socially efficient goal for principals and teachers. Further, the variation in proficiency rates among schools reflects, in large part, interschool differences in student background characteristics. The designers of accountability systems must move away from systems designed around a one-size-fits-all standard and begin designing systems that organize and promote competition among schools. Well-organized competition among schools is the best vehicle for making sure that schools use public funds efficiently. If education officials pursue this paradigm, they must develop relative performance measures that assess the outcomes of these contests while making reasonable allowance for differences in student populations served by public schools. I will discuss a method for deriving context-specific measures of school performance. A percentile performance index tells public officials how often the students in a particular school or classroom perform better than students in other schools who began the year in similar circumstances with respect to their prior achievements, the compositions of their classmates, and their family backgrounds. This index of relative performance provides the information policymakers need to make preliminary judgments concerning when to reorganize a given school and give a new staff the opportunity to prove they can do better.