The increasing complexity of policy problems, coupled with the political desire to base new policies on the foundation of firm evidence, has accelerated the development of policy assessment tools. These range from complex computer models and cost benefit analysis through simple checklists and decision trees. In the last decade, many governments have established formal policy assessment systems to harness these tools in order to facilitate more evidence-based policy making. These tools are potentially widely available, but to what extent are they used by policy makers and what becomes of the evidence that they generate? This paper addresses these questions by studying the empirical patterns of tool use across 37 cases in three European countries and the European Commission. It uses a simple classification of tools into advanced, formal and simple types. It finds that even when tools are embedded in policy assessment systems, their use is differentiated and on the whole very limited, in particular when it comes to more advanced tools. It then explores these patterns from contrasting theoretical perspectives to shed light on why, when and how different policy assessment tools are used in the policy process.