Knowledge creation is a collective and social activity, and a large body of research has established that knowledge creation by researchers (knowledge workers) is influenced by their direct exchange partners. We examine why knowledge worker ego networks are structured as they are, but also why and how knowledge worker networks change over time. We examine two changes to knowledge workers ego networks: the addition of new direct exchange partners and the deletion of existing direct exchange partners. Our study offers important evidence that two network dimensions (tie strength between ego and his or her alters, and the level of connectivity among egos alters) provide distinctive insight into how networks change. Importantly, our study provides evidence that three critical components of the knowledge creation processaccess to diverse and redundant resources, shared experiences and tacit mutual understanding to develop the resources, and previous knowledge creationsact as causal mechanisms behind network changes in subsequent periods. Our study is at the individual level as we study biomedical research scientists and their direct exchange partnersthe others to whom they are directly connected through coauthorship.