This paper proposes a theory of education curriculum and analyzes its distributional impact on student learning outcomes. Different curricula represent horizontal differentiation in the education technology, thus a curriculum change has distributional effects across students. We test the model using the quasi-natural experiment of the G8 reform in Germany. We find evidence of heterogeneous reform effects consistent with our theory. While the reform improves student test scores on average, such benefits are more pronounced for well-prepared students. In contrast, less-prepared students do not benefit from the reform.