Many philosophers take mind-independence to be criterial for realism about kinds. This is problematic when it comes to psychological and social kinds, which are unavoidably mind-dependent. But reflection on the case of artificial or synthetic kinds (e.g. synthetic chemicals, genetically modified organisms) shows that the criterion of mind-independence needs to be qualified in certain ways. However, I argue that none of the usual variants on the criterion of mind-dependence is capable of distinguishing real or natural kinds from non-real kinds. Although there is a way of modifying the criterion of mind-independence in such a way as to rule in artificial kinds but rule out psychological and social kinds, this does not make the latter non-real. I conclude by proposing a different way of distinguishing real from non-real kinds, which does not involve mind-independence and does not necessarily exclude psychological and social kinds.