Most integrated assessments conducted on complex environmental issues, such as climate change, failed to fully meet the needs of the decision makers they are intended to serve. This observation has been recognized by many practitioners over the past two decades and also has been documented by social learning research. When researchers apply assessment models as mechanistic predictive tools, they risk producing results that are not sufficiently relevant and prone to misuse in decision making. Policy-relevant assessment is more of an organic process involving researchers and the issue's stakeholders in an interactive exercise that builds understanding rather than defines truth. The chapter examines some of the shortcomings of past assessment efforts, identifies possible solutions, and provides examples of recent work on climate change and air quality planning where new approaches have been applied successfully.