<p>Kripke convincingly argued that proper names in natural language are rigid designators, and observing that natural kind terms resemble proper names in important respects, he suggested that they are also rigid designators. But it is not immediately clear what exactly this means, for while rigidity as defined by Kripke is primarily a property of singular terms, some of the natural kind terms discussed by him are general terms functioning as predicates. In this paper, I propose a definition of rigidity for predicative general terms, motivated by Putnam’s thesis that the rigidity of the term ‘water’ is a consequence of its indexicality. I argue that this definition satisfies the three desiderata put forward by Soames for an adequate notion of general term rigidity: (i) that it be a natural extension of the notion of singular term rigidity; (ii) that the class of rigid general terms roughly coincide with the class of natural kind terms; (iii) that it play the right role in explaining the necessity of true <i>a posteriori</i> theoretical identifications. I also compare this notion of rigidity with a certain notion proposed by Nimtz as an alternative to general term rigidity, explaining how they differ in crucial respects. On the conception of rigidity I advocate, rigid general terms can be characterized as terms that are semantically associated with what Locke called real essences.</p>