abstract The notion of a ‘community-of-practice’ (CmP) has become a highly influential way of conceptualizing how decentralized sub-units or groups within firms or organizations operate. CmPs refer to ‘tightly knit’ groups that have been practising together long enough to develop into a cohesive community with relationships of mutuality and shared understandings. The CmP notion, however, does not fit squarely with how temporary organizations or project organizations operate. Typically these kinds of groups consist of diversely skilled individuals, most of whom have not met before, who have to solve a problem or carry out a pre-specified task within tightly set limits as to time and costs. As a result they tend to become less well-developed groups, operating on a minimal basis of shared knowledge and understandings. Such a group, I suggest, constitutes a ‘collectivity-of-practice’ (ClP). Mirroring the above distinctions, two ideal-type notions of epistemology are developed. The one inspired from the CmP literature is discussed in a ‘knowledge community’ terminology, whereas the one associated with ClPs is conceived of as a ‘knowledge collectivity’. Finally, I outline some new options for organizational analysis made possible by recognizing these as two different and complementary notions.