In research on industrial districts, network relations among firms and organizations are considered critical for the development and diffusion of new knowledge, with implications for innovation and regional development. A review of this literature, however, suggests that many investigators are surprisingly silent, conceptually and empirically, about the structure of innovative networks, although they often couch their arguments implicitly in structural language. The survey of a sample of 64 empirical industrial district studies shows that few researchers have measured and analysed network structures, and none have studied performance outcomes of variations in network structure. An evolutionary perspective is used to propose a research agenda for a better understanding of the relationship between network structure and innovation.