Despite growing interest in the impact of AIDS on educational outcomes, it is unclear how or why orphans might be educationally disadvantaged compared with unorphaned children and how to intervene to mitigate this disadvantage. This research attempts to determine the impact of parental death on educational outcomes; the mechanisms involved; and whether social grants mitigate any observed impact. This thesis analyses educational outcomes of orphans in South Africa using data on 1635 school-age children from the KwaZulu-Natal Iii -ome Dynamics St»d), (KIDS), a panel of households that has been surveyed in 1993,1998 and 2004. All three waves of KIDS have collected demographic and household expenditure data. In addition, the 2004 wave collected detailed information on children's schooling. This dataset has been linked to official statistics on schools to control for school effects. Regression modelling is used to control for confounding factors and identify causal pathways. More than a third of children aged 7-20 in the study are orphans. Paternal and dual orphans tend to live in poorer households than other children but also tend to come from urban areas and to have relatively educated parents. Death of a parent more than doubles the risk of late enrolment in school for both boys and girls. Paternal orphanhood is significantly associated for girls with late or noncompletion of primary school, grade repetition, and dropout; and for teenage boys with poor attendance. The poor outcomes of paternal orphans persist when controlling for the poverty and other characteristics of the household before the father's death. Important causal mechanisms include difficulties with meeting the costs of schooling and higher levels of teenage pregnancy. Cash grants reduce the educational disadvantage of poor children but do not significantly offset the specific adverse effects of orphanhood on educational outcomes.