This study estimates mean and distributional effects of early between-school ability tracking on student achievement. For identification, I exploit heterogeneity in tracking regimes between German federal states. After comprehensive primary school, about 40% of students are selected for the academic track and taught in separate schools in all states. The remaining students, however, are either taught comprehensively or further tracked into two different school forms depending on the state. I estimate the effects of this tracking on students' mathematics and reading test scores with a difference-in-difference-in-differences estimator to eliminate unobserved heterogeneity in achievement levels and trends between states. I find substantial achievement gains from comprehensive versus tracked schooling at ages 10-13. These average effects are almost entirely driven by low-achievers. I do not find evidence for negative effects of comprehensive schooling on the achievement of higher performing students. My results show that decreasing the degree of tracking in early secondary school can reduce inequality while increasing the efficiency of educational production.