The swift rise of policy appraisal in recent years – to the status of legal requirement in many countries - has been mirrored by development of many support tools such as environmental models. However, there is a widely- observed gap between extensive supply of such tools, and patchy demand for them; their use is influenced by many technical, procedural, conceptual and political factors. How and to what extent can the relationship between appraisal tools and policy-making be developed, particularly the type of expertise required for addressing complex policy problems such as climate change? This paper addresses this question within the wider concept of science-policy interaction as a fluid and porous interface, and also as a shared, multi-actor process of addressing policy problems. The paper employs a knowledge-brokerage (KB) approach, where the linear model of ‘speaking truth to power’ is challenged by a more reflexive approach to the interaction. To do this, and to reflect the many context-specific forms of the science-policy interface, we focus on case studies of six different policy problems at different decision making levels and jurisdictions. We assess the most appropriate KB strategy in each case, and critically evaluate the KB approach – how it worked, what factors influenced it and how effective it was. Using the preliminary case study results, we describe a preliminary typology for different 'modes' of SPI operation, and discuss how institutional setting affects the process, governance and success of knowledge brokerage. Work is ongoing, but initial results show that the 'test case' concept acts as a flexible conceptual and practical guidance for researchers in science policy interactions in policy appraisal processes, and can help facilitate the relationship between scientists and policy makers. The approach yields conceptual learning about the science-policy interface, and reveals different actors' conceptual models of knowledge production and application.