Many South African school children are known to travel fairly long distances to school each day, in pursuit of the best possible educational opportunities in a schooling system that is known to vary greatly in quality. This thesis documents the dimensions and determinants of the daily, education-related travel of primary school aged children in Johannesburg-Soweto, South Africa. It uses data on a sample of 1428 children drawn from the Birth to Twenty cohort study to provide the first population-based data on the extent of learner mobility in contemporary urban South Africa. Learner mobility is measured in three different ways: firstly by the straight line distance between a child‘s home and his or her school; secondly by whether the child‘s school falls into the same geographical area as his or her home; and thirdly by whether the child attends his or her nearest, grade-appropriate school. The thesis provides clear evidence for extensive mobility using all three of these approaches to measurement. Over 25% of children were found to be travelling more than 5km each way to school and back on a daily basis. Almost 60% of children attended a school outside of the Census 2001 Sub-Place (roughly equivalent to a suburb) in which they lived, and fewer than 20% of children attended the grade-appropriate school nearest to their home. Counter to expectations, these figures were fairly stable over time, suggesting that educational mobility does not increase substantially as children age or transition to high school. Mobile children attended significantly more well-resourced and well-performing schools than their non-mobile peers, and the quality of schools attended increased with distance travelled. This substantiates the assumption that children and families make use of educational mobility to improve the quality of education that they are able to access. The analyses presented in the thesis suggest that two distinct patterns of mobility, with different determinants, are in use in the Johannesburg-Soweto area. The first relates primarily to travel from townships to historically advantaged schools in suburban Johannesburg, and typically requires substantial economic investment and extensive parental involvement. The second form of mobility operates at a more local level, and relates to children and families making choices between a number of relatively local schools. This form of mobility is less resource intensive. Children engaging in the first form of mobility were more likely to attend a particularly advantaged school, and to have a well-educated mother. By contrast, children engaged in the second form of mobility were more likely to live in a disadvantaged area, and come from households with moderate SES levels. iv The findings of this thesis provide important insights into the nature of school choice in South Africa, which have implications for educational policy, and the understanding of the nature of urban poverty as experienced by South African children. They also contribute to the international school choice literature, by providing novel information about the implications of relatively unregulated school choice for educational inequality and segregation in the South African context.