Indicators of school-level achievement, such as the percentage of students who are proficient in a particular content area, are subject to random year-to-year variation in much the same way that the results of an opinion poll will vary from one random sample to another. This random variation, which is more pronounced for a small school, should be taken into account by education officials when evaluating school progress in a policy climate of high stakes. To do otherwise is to unnecessarily risk the false identification of a failing school. In this monograph, I describe the application of confidence intervals to the evaluation of "adequate yearly progress" for No Child Left Behind (NCLB). Throughout, I demonstrate the particular relevance of confidence intervals for small schools. Upon completion, readers will understand why 27 states included confidence intervals in their NCLB accountability plans (and perhaps wonder why the remaining states did not). (Contains 13 footnotes, 5 figures and 1 table.)