This thesis aims to answer three main questions ; how can one explain andinterpret the structure of an innovation network, are there positions in a network which allowfor an increased performance for firms and finally, are there network structures which favourinnovation ? In order to answer these questions, the thesis is organised in three parts.The first part presents, in a first chapter, an analytical review of the literature followed by achapter presenting the theory behind one of the network analysis methods: ExponentialRandom Graph Models (ERGM).The second part of the thesis presents three empirical analyses. The first empirical chapteranalyses the impact of the life-cycle of the technology on the structural dynamics of thecollaboration network for Structural Composite Materials. The following two chapters focuson two sectors, the aerospace and biotech sector. The aim of these chapters is to analyse thestructural dynamics of collaboration networks as well as identifying a link between networkposition and firm performance.The third and final part of this thesis searches for network structures which might favourinnovation. An Agent-Based Model is used to answer this final question.