This thesis explores the use of scientific knowledge in policy-making. It is widely recognised that use of natural resources, such as water, is unsustainable across much of the world. In response, there is a desire for and expectation that science should be used to help improve natural resource management and contribute to the pursuit of sustainability. This research is centred on a comparative case study analysis of the use of science in water resource management policies for Southeast Queensland, Australia. A framework model of the interface between science and policy is developed and applied to the case studies and a range of different uses for science in policy-making are synthesised in the analysis. The relationship between these uses and the need for boundary work to make the science effective is analysed. The findings explore the key factors that influence how science can be used to create effective, efficient and appropriate policies. The research revealed that science can be used in a number of different ways, from simply providing knowledge through to being used in a co-learning process to create a shared understanding. The results show that co-learning is likely to be the most effective use of science, especially when the science is uncertain, but this use requires strong social capital and collaborative governance arrangements. More direct knowledge provision is simpler to implement, but is likely to require shared values and more certain science. In controversial policy issues, where there are conflicting values, science is likely to be used in advocacy and this limits its ability to directly influence outcomes. Hence, the use of science is strongly influenced by the policy context and governance arrangements, but generally strong social capital is likely to support the most effective use of science. This adds further support to calls for stronger social capital and more collaborative governance in sustainability and natural resource management policy.