This paper presents an approach to mathematical terms similar to the approach developed by Kripke in order to deal with natural kind terms; in fact, I argue that mathematical terms are natural kind terms in the sense of Kripke. Thus, I suggest that, from a semantic perspective, such terms should be seen as primarily referring terms, and, from a metaphysical perspective, that good definitions of such terms should embody structural information about the exemplars from the kind in question.