Educational tracking is a very controversial issue in education. The tracking debate is about the virtues of uniformity and vertical differentiation in the curriculum and teaching. The pro-tracking group claims that curriculum and teaching better aimed at children's varied interest and skills will foster learning efficacy. The anti-tracking group claims that tracking systems are inefficient and unfair because they hinder learning and distribute learning inequitably. In this paper we provide a detailed within-country analysis of a specific educational system with a long history of early educational tracking between schools, namely the Flemish secondary school system in Belgium. This is interesting place to look because it provides a remarkable mix of excellence and inequality. Indeed the Flemish school system is repeatedly one of the best performer in the international harmonized PISA tests in math, science and reading; whereas it produces some of the most unequal distributions of learning between schools and students. Combining evidence from the PISA 2006 data set at the student and school level with recent statistical methods, we show first the dramatic impact of tracking on social segregation; and then, the impact of social segregation on equality of educational opportunity (adequately measured). It is shown that tracking, via social segregation, has a major effect on inequality of opportunity. Children of different economic classes will have different access to knowledge.