Previous international work has shown that clustering pupils with similar characteristics in particular schools yields no clear academic benefit, and can be disadvantageous both socially and personally. Understanding how and why this clustering happens, and how it may be reduced, is therefore important for policy. Yet previous work has tended to focus on only one kind of clustering at a time. In the USA, for example, black:white segregation of pupils has been the key issue. In the UK, and across Europe, the focus has been on social background, especially on the segregation from their peers of pupils living in poverty. With access to high-quality national data sets in England, it is now possible to track the between-school segregation of numerous pupil characteristics over a 14-year period. This paper uses school-level figures for all state-funded secondary school pupils in England from 1996 to 2009, including free school meals, special needs, ethnicity, and first language. The paper investigates, more fully than has been attempted before, whether segregation in terms of these indicators has a common pattern. It presents evidence for the existence of at least three different processes of between-school segregation over this period, and proposes a possible determinant for each, and some future work that could be done to clarify the situation. The paper therefore provides an important corrective to previous work that sought explanation for only one type of segregation. In the allocation of school places, pupils are being clustered in several distinct ways. The standard previous explanations for pupil clustering such as selection by ability or housing, faith-based enrolment, and increased parental choice, apply only to some of these forms of segregation. (Contains 5 tables.)