Abstract and Key Results * MNEs are moving away from a 'centralised hub' to a 'multi-hub' network of R&D units. Using evidence from European pharmaceutical MNEs, this study analyses the challenges associated with promoting and integrating knowledge flows in multi-hub R&D organisational structures. * While these new structures provide greater potential for cross-fertilization of technologies and access to location-specific competences, firms also need to overcome greater levels of inter-unit geographical, organisational and technological distance. Firms also suffer from organisational inertia, which further hinders lateral communication and inter-unit knowledge transfer. * There are important variations in the way in which integrated network structures have been implemented, but in general, these new structures have increased the need for coordination mechanisms, but ironically most companies have reduced or eliminated this 'traditional' headquarters function. * While socialization mechanisms help to overcome some of these bottlenecks, there remain a number of obstacles in optimising knowledge flows in physically and technologically dispersed R&D facilities. Key Words R&D, Multinational Enterprise, Organizational Design Introduction In a relatively short span of three decades, the extent, spread, motivation, location and nature of the overseas R&D activities of multinational enterprises (MNEs) has become incredibly complex. Some of these changes reflect the increasingly complex nature of MNE activities, as the nature of headquarters--subsidiary relationships have been re-organised away from an ethnocentric, home country dominated structure, to a more widely distributed and complex network of knowledge flows between subsidiaries and headquarters in several locations. MNEs are moving away from a 'centralised hub' to a multi-hub, 'integrated network' of R&D units which contribute to the creation of new technological assets building on host-location specific knowledge assets. This is reflected in the growing geographical spread of MNE's centres of excellence (see for example Holm/Pedersen 2000), the growing phenomenon of reverse technology transfer (e.g., Hakanson/Nobel 2000, 2001, Frost/ Zhou 2005), and the adoption of new R&D organisational structures (e.g., Chiesa 1996, Gassmann/von Zedtwitz 1999, von Zedtwitz/Gassmann 2002). Although these new structures provide greater opportunities for cross-fertilization of technologies and access to location-specific, MNEs are just beginning to realize the costs, constraints and challenges associated with efficiently managing geographically dispersed activities. Some recent studies have suggested mechanisms to improve the efficiency of implementing a multi-hub approach (e.g., Nobel/Birkinshaw 1998, Reger 2004). However, while much of this current research has discussed the superiority of the multi-hub approach over the centralised hub model (and variants such as the centres of excellence approach), it tends to pay less attention to the challenges in achieving the increased knowledge flows and cross-disciplinary and cross-border networking abilities that are essential for its successful implementation. There are indeed a number of barriers to the internal knowledge diffusion process connected to inter-unit geographical, organisational and technological distance and also to the motivational disposition of both the sender and the receiver units (see Kogut/Zander 1993, Szulanski 1996, Gupta/Govindarajan 2000). In addition, firms experience a certain amount of organisational inertia, in that despite new structures being implemented, knowledge flows and coordination de facto continue to sub-optimally follow the same patterns associated with the old structure. This paper aims to analyse the managerial challenges in achieving knowledge transfer inside multi-hub integrated R&D networks and in overcoming organisational inertia. …