Governments in OECD countries are turning more and more towards creating networked entities as a means of organising cross-sector and multidisciplinary research. Yet, there is little discussion of how such networks operate and how they differ in evaluation terms from other research entities (individuals and organisations). This particularly relates to the policy objectives of networks. In this paper, we use the literature on evaluation, impact and value as a lens through which to focus on the nature and benefits of formal research networks. This paper seeks to refine our concepts of research networks and, in defining the concept of formal research networks, to map the policy issues in evaluating networks. We argue that, to do this, it is important that two extant literatures (stakeholder theory and organisational environments) be introduced into the analysis of network operations. We focus particularly on the significance of environmental complexity for network evaluation.