As a result of the release of the Burdekin Report in 1989, the Commonwealth Government developed a series of initiatives in the August budget of that year, including an educational program, the Students at Risk Program. In Victoria, the Ministry for Education developed and implemented seven Projects within the Students at Risk Program, including one on Homelessness. This thesis reports a two years study of the development and implementation of that Homelessness Project. The study confirms that schools can successfully assist homeless and 'at risk' students to remain in school. It shows that there was clear mismatch between the rhetoric of the policy and the nature of practice in each of the schools. This mismatch is attributed to a range of interrelated influences. These influences included the complex nature of the homelessness problem, different approaches to the resolution of the problem in each school, a continual breakdown in social relations between participants throughout implementation, continual recontextualising of the policy by staff in each school, repeated contestation of official directives, the radicalisation of the school principals involved, and the iterative nature of the policy process. The study of implementation of the Homelessness Project has highlighted the need for policy directed to 'at risk' students in schools to be developed and implemented in a co-operative, democratic way to avoid rejection, reinterpretation and manipulation of policy at the school level. The debate over the nature of policy is far from resolved and conceptualisations about the nature and role of the state, and the nature of power relations in the policy process, are more complex than is generally asserted in the literature. This thesis contributes to the debate and suggests that the state can and does affect the nature of policy acceptance at the micro level, but that this is often countered by the strength of institutionalised practice in schools. In conclusion, therefore, whilst there is evidence that both the state and the school affect implementation, the effects are neither homogeneous nor continuous. Recognition of variations in the degree of influence of the state and schools might eventually lead to a fuller understanding of the nature of policy and allow more larger scale theorising to be derived.