A considerable body of literature has investigated how environmental exposures affect health through various pathways. These studies have generally adopted a common approach to define environmental exposures, focusing on the local residential environment, using census tracts or postcodes to delimit exposures. However, use of such administrative units may not be appropriate to evaluate contextual effets on health because they are generally not a 'true' representation of the environments to which individuals are exposed. Recent work has suggested that advances may be made if an activity-space approach is adopted. The present paper investigates how various disciplines may contribute to the refinement of the concept of activity space for use in health research. In particular we draw on seminal work in time geography, which provides a framework to describe individual behavior in space and time, and can help the conceptualization of activity space. In addition we review work in environmental psychology and social networks research, which provides insights on how people and places interact and offers new theories for improving the spatial definition of contextual exposures.