The article highlights two contrasting ways in which social theorists have been trying to define the ontological boundaries of sociology since the early days of the discipline. Some (e.g. Durkheim, Weber, and critical realists) have attempted to demarcate social reality as a causally autonomous and qualitatively distinct realm in a segmented/stratified universe. Others (e.g. Tarde, Spencer, Luhmann, sociobiologists, and actor-network theorists) have postulated a more open (or flat) ontological space and blurred such demarcations by either rejecting the causal autonomy of sociological phenomena, or their qualitative distinctiveness, or both. So far, there has been little convergence between these two orientations since according to the former, the opening of the boundaries is likely to give way to reductionist conceptions of society, whereas the latter tends to associate rigid boundaries with essentialism. Through a close examination of these opposing orientations, the article aims to shed light on current ontological dilemmas of sociology.