The view that social kinds (e.g., money, migrant, marriage) are mind-dependent is a prominent one in the social ontology literature. However, in addition to the claim that social kinds are mind-dependent, it is often asserted that social kinds are not real because they are mind-dependent. Call this view social kind anti-realism. To defend their view, social kind anti-realists must accomplish two tasks. First, they must identify a dependence relation that obtains between social kinds and our mental states. Call this the Dependence Task. Second, they must show that social kinds are not real because they are mind-dependent. Call this the Anti-Realist Task. In this paper, I consider several different ways of defining the relation that is supposed to obtain between social kinds and our mental states. With respect to each relation, I argue that either it fails to accomplish the Dependence Task, or it fails to accomplish the Anti-Realist Task. As such, anyone who wishes to defend social kind anti-realism must provide an alternative explanation of how social kinds depend on our mental states in a way that impugns their reality. In the absence of such an explanation, there is no reason to endorse social kind anti-realism.