The focus of this paper is cognitive convergence in a globally distributed team (GDT), defined as the process by which cognitive structures of distributed team members gradually become more similar over time. To explore the convergence process, we employed a longitudinal, ethnographic research strategy that allowed us to follow a naturally occurring GDT over a 14-month period, producing a rich case study portraying factors and processes that influence convergence. Confirming previous studies, we find that increases in shared cognition alone are not sufficient to account for performance gains on a GDT. Rather, it may be necessary not only to increase the sharing of cognition, but also to reverse a pattern of increasing divergence that can result from rejection of key knowledge domains. We also found that several factors influence the process of cognitive convergence beyond direct knowledge sharing. These include: separate but parallel or similar learning experiences in a common context; the surfacing of hidden knowledge at remote sites by third-party mediators or knowledge brokers; and shifts in agent self-interest that motivate collaboration and trigger the negotiation of task interdependence. Also relevant to cognitive convergence on a GDT is the geographical distribution pattern of people and resources on the ground, and the different ways in which leaders exploit the historical, cultural and linguistic dimensions of such distribution to further their own political agendas. Several propositions related to these observations are suggested. We conclude that GDTs can be effective in bringing together divergent points of view to yield new organizational capabilities, but such benefits require that leaders and members recognize early and explicitly the existence and validity of their differences. Copyright © 2004 John Wiley & Sons, Ltd.