People often maintain stereotypes despite obvious counterexamples. Several philosophers have recently argued that this persistence is due to stereotypes being statistical rather than universal generalizations and therefore compatible with counterexamples. In this paper, we argue that this is not the only reason why stereotypes are so persistent. Many stereotypes also involve causal-explanatory beliefs of the following type: ‘There is a causal property P entangled with social category K that causes F when it operates without interference’. Such causal-explanatory beliefs allow for counterexamples because they entail generalizations that are limited in scope; they purport to describe only those members of a social group K in which the presumed causal property P operates without interference. Our analysis provides a more complete account of why stereotypes are so persistent, why they are morally objectionable, and why addressing them effectively requires more than simply pointing out exceptions.