Science–policy-interfaces (SPIs) are expected to go beyond the linear model of scientific policy advice through creating spaces for exchange and dialogue between ‘policy’ and ‘knowledge’. Given that most environmental issues require inter- and transdisciplinary approaches, SPIs must take into account a variety of knowledge types, views and interests of scientists, policymakers and other decision makers. Moreover, acceptance and durability of SPIs depend largely on their perceived legitimacy and the credibility of their knowledge-gathering processes, providing additional challenges for their internal organisation. As the interplay between different knowledge types and decision making is far from neutral, a reflexive approach is required in the design of an SPI so that it is capable of learning from past experiences. The aim of this article is to discuss which governance arrangements could best support the development of an effective and legitimate SPI for European biodiversity politics. We analyse different options for facilitating the implementation of a ‘Network of Knowledge’ approach. This approach has been developed to improve the interface between diverse knowledge-holder communities and decision making processes for biodiversity and ecosystem services—a field where multi-scalar and multi-dimensional problems arise. In this article, we develop and discuss two stylized extreme governance models as our starting point: an `informal network model´, which almost entirely depends on the dedication of individuals, versus a more formalized `platform model´, predominantly based on the needs and interests of the organisations involved. We discuss the pros and cons of each of these models in reaching their objectives and in developing sound governing processes for a ‘Network of Knowledge’. From this discussion, we derive a recommended design for the reflexive governance of such a network in the context of the European Union and finish by discussing some more general lessons learnt.