Crime pattern theory is a central framework within environmental criminology, providing a means to understand how the spatial structure of human activities can shape the distribution of crime in urban areas. An important idea is the role played by shared activity nodes: locations that feature in the routine activities of many people, and therefore give rise to many interpersonal interactions. Such locations are proposed to play a prominent role in shaping patterns of crime and bring about significant local concentrations of victimisation. Despite the intuitive appeal of these mechanisms, however, their definition is remarkably broad: the term ‘shared’, for example, encompasses a wide range of scenarios, from local shops visited by small numbers of people to transit stations frequented by many. It is not clear at what point in this spectrum (if at all) the behaviour becomes sufficient to generate the characteristic concentrations predicted by crime pattern theory. In this chapter, we employ a computational model of interpersonal victimisation to investigate this issue by examining varying realisations of routine spatial behaviour in an abstract urban environment. Using this model, we examine the extent to which the overlap of activity spaces is associated with crime concentrations at a local level, including the contribution of guardianship. We also analyse the resulting patterns at a range of scales, finding that the effects of shared activity nodes may extend much further than their immediate vicinity, and that their macro-level effect may, counter-intuitively, be a crime-reducing one.