This dissertation presents a case study of the evolution of water supply policy in Zambia, applying Rhodes and Marsh's 'policy network theoretical approach. Since this approach was developed in northern, industrialized countries, and has not been tested through application in developing countries, my research question is: can the policy network approach explain the policy process for peri-urban water supply in Lusaka, Zambia during the period 1991 to 2008? To answer this question and characterize the governance of the water sector, I undertook in-depth interviews with almost 50 respondents and reviewed relevant documents. After years of policy stagnation, the newly-elected Zambian government in 1991 embarked on a series of reforms, which divided responsibility for water supply between several ministries and levels of government, an inter-ministerial policy agency and commercial utilities. International actors brought in new funding and policy ideas, and the period of the 1990s was marked by a multiplication of models of water supply. Learning was adapted from both international and national experiences of peri-urban water supply. Two main policy experiments emerged, the commercial model and the community participation model, which developed relatively independently. In the latter years of the study period, due to the growing resource capacity of the Zambian government and the donors' agenda of aid harmonization, these two models became more integrated. The policy network approach did foresee this type of shift, from a phase of change with a diversity of actors and models, to a phase of policy consolidation and coordination. The network framework draws attention to an important set of policy dynamics. In its application to a developing country setting, however, the approach doesn't prepare the researcher for the marked differential between a weak state and strong donors, and generally hasn't worked out the complexities of relationships between international, local and community based organizations.