Our education system is in desperate need of innovation. Despite radical advances in nearly every other sector, public school students continue to attend school in the same buildings and according to the same schedule as students did more than a hundred years ago, and performance is either stagnant or worsening. One of the most important innovations in the past halfcentury is the emergence of charter schools, which, when first introduced in 1991, came with two distinct promises: to serve as an escape hatch for students in failing schools, and to create and incubate new educational practices. We examine charter schools across the quality spectrum in order to learn which practices separate high-achieving from low-achieving schools. An expansive data collection and analysis project in New York City charter schools yielded an index of five educational practices that explains nearly half of the difference between high- and low-performing schools. We then draw on preliminary evidence from demonstration projects in Houston and Denver and find the effects on student achievement to be strikingly similar to those of many high-performing charter schools and networks. The magnitude of the problems in our education system is enormous, but this preliminary evidence points to a path forward to save the 3 million students in our nation’s worst-performing schools, for a price of about $6 billion, or less than $2,000 per student.