Historically, archaeologists working on non-state societies have tended to interpret religion and large-scale religious architecture as necessarily integrative, that is, as naturalizing the social order or producing an abiding sense of community. I argue here that this focus on integration has limited our ability to understand how and why religion changed through time and how religion may have been a driver of social change. We will benefit from considering the political dimensions of religious practice in non-state societies as much as in more "complex" settings. This study explores the articulation of religious practice and religious architecture with social and spatial boundaries in the prehispanic U.S. Southwest. In particular, I examine variability and change in rectangular great kivas—large, semi-subterranean religious structures—in west-central New Mexico and east-central Arizona between 1000 and 1400 CE in relationship to socially diverse contexts that might be viewed as borderlands or frontiers. The study pulls together two broads strands of research. The first is an examination of the unusual great kiva at the thirteenth-century CE Fornholt site (LA 164471) near Mule Creek, New Mexico, in relation to the broader history of the surrounding Upper Gila area. This portion of the research is based on two seasons of excavation at Fornholt and on an examination of records and ceramic collections from the Upper Gila. I suggest that the Upper Gila may be considered a borderland or frontier through time and that viewing Fornholt as a borderland site sheds light on the site's material culture, including its great kiva. The second strand of research is a comparison of great kiva architecture and assemblages across the larger study area based on the examination of museum collections and the aggregation of published and unpublished architectural data. The broader study demonstrates that, while these great kivas make up a coherent tradition and fit within the larger world of southwestern religion, great kivas in borderland contexts show experimentation and change in ways that more centrally located great kivas do not. I argue that this diversity can be viewed in light of the negotiation of social boundaries in borderland contexts, including the role of great kivas as political venues or contested spaces.