What is the long-run impact of high-stakes testing on students’ life chances? The past decade of research on school accountability has found that high-stakes testing increases student achievement, and also that it leads to behavioral distortions such as teaching to the test, manipulation of the exam-eligible pool of students, and teacher cheating. Neither of these results tells us whether school accountability improves student welfare. We investigate the long-run impact of accountability pressure on students in Texas public schools. Texas was one of the first states to implement an NCLB-style accountability system. Beginning in 1993, schools were assigned performance grades based on the percent of students who passed state exams in math and reading. We first show that high schools with low initial scores under the low-stakes regime made relatively larger gains on the state exams in the first few years of accountability - the so-called “Texas miracle.” We then demonstrate that students in low-performing schools with particularly large test score gains also saw improvements in high school graduation, college attendance and completion, and earnings. Our preliminary results indicate that school accountability increased attainment and earnings for students in low-performing schools, but not for students in the middle or at the top of the initial distribution.