In this PhD project we assume that there is no unique, ideal science-policy interface but that the performance of these interfaces (in terms of producing policy-relevant knowledge and mutual learning) depends on the policy activity at issue. Especially in the case of policy development and management of controversial issues we expect a high risk of strategic use of knowledge and report wars. In these cases scientists may have to play another role than in the case of monitoring and maintenance. In this project the policy activity-specific requirements for science-policy interfaces and how these can be met are addressed.