Grouping students into classes by ability on a subject-by-subject basis, also known as setting, is a practice common in many educational systems. Arguments made for its effectiveness include the advantage for teachers of teaching students of similar ability which makes it easier to pitch lessons at the right level, and the fact that setting makes it easier to provide experiences of success for all students, as well as greater opportunities for peer learning as pupils do not experience great disparities between each other. On the other hand, critics point to the possibility of harming a student's self-concept when that student is put into a lower set, to the fact that it is extremely difficult for teachers to have high expectations low set classes, and the loss of opportunities for lower achieving peers to be peer taught by higher achievers in the subject (Muijs & Reynolds, 2001).Research has shown that students in lower streams are often taught an 'impoverished' curriculum, which does not attempt to engage student's thinking or understanding and is limited to simple drill and practice exercises (Hansell & Karweit, 1983). While in theory designed to teach students at their level, this practice can further widen the gap between the high and low streams and fails to give all students access to a rich, full curriculum, thus creating equity issues.An important issue is the way in which setting decisions are made. While educators and policymakers favouring setting claim that ability or achievement are the sole criteria used, critics counter that rather than being determined by achievement or ability, there is a tendency for factors such as pupils' socio-economic status background, gender and ethnicity to influence teachers' setting decisions, presenting clear equity problems (Wright-Castro, Ramirez & Duran, 2003).This paper aims to examine the extent to which factors other than prior achievement influence setting decisions in England, using an equity framework.In order to test the extent to which achievement or background factors determine setting decisions, a large scale survey was carried out into setting practices in 12 Local Authorities in England. A survey questionnaire was sent to a random sample of 400 primary and secondary schools in these authorities, asking schools about setting practices, criteria and motives in English and Maths. 166 responses were received. Respondents were also asked to provide data on which pupils were in which sets for these subjects giving their Unique Pupil Number. This data will be linked to data from the National Pupil Database held by the DfES, in which each pupils' prior achievement in national tests, gender, free school meal eligibility, post code, ethnicity and other background factors are collected. Multilevel logistic and multinomial regression models will be used to test which factors actually predict membership of higher and lower sets in these subjects, in order to determine empirically what factors are actually related to setting. This will then be compared to data from the survey reporting educators' stated criteria. The use of multilevel models will allow us to look at the extent to which predictors differ between schools, and will represent the first quantitative study to look at between-school variance in this area.The study will determine what variables predict mebership of different ability sets, exploring differences at school level as well as looking at pupil level predictors. This will enable us to determine the extent to which these decisions are based on ability or other factors. The paper will therefore make a contribution to both theory and practice in education. Hansell, N & Karweit, S. (1983). Friends in School. Washington: Academic Press. Muijs, D. & Reynolds, D. (2001). Effective Teaching. Evidence and Practice. London: Paul Chapman. Wright-Castro, R., Ramirez, R. & Duran, R. (2003). Latino 6th Grade Student Perceptions of School Sorting Practices. Paper presented at the Annual Meeting of the American Educational Research Association, Chicago, Ill, 23 April 2003