Getting evidence into policy is notoriously difficult. In this empirical case study we used document analysis and key informant interviews to explore the Australian federal government's policy to implement a national bowel cancer screening programme, and the role of evidence in this policy. Our analysis revealed a range of institutional limitations at three levels of national government: within the health department, between government departments, and across the whole of government. These limitations were amplified by the pressures of the 2004 Australian federal election campaign. Traditional knowledge utilisation approaches, which rely principally on voluntarist strategies and focus on the individual, rather than the institutional level, are often insufficient to ensure evidence-based implementation. We propose three alternative models, based on deliberative strategies which have been shown to work in other settings: review of the evidence by a select group of experts whose independence is enshrined in legislation and whose imprimatur is required before policy can proceed; use of an advisory group of experts who consult widely with stakeholders and publish their review findings; or public discussion of the evidence by the media and community groups who act as more direct conduits to the decision-makers than researchers. Such deliberative models could help overcome the limitations on the use of evidence by embedding public review of evidence as the first step in the institutional decision-making processes.