This paper studies Robert Boyle's account of kinds and classification. A number of commentators have argued that, for Boyle, classifications are inevitably the product of conventions. Others have challenged this reading, arguing that, according to Boyle, the corpuscular makeup of bodies gives rise to hard-edged natural kinds and classes. We argue that Boyle's position is more complicated than the available realist and conventionalist readings acknowledge. We argue that, according to Boyle, the individuation of kinds was to some degree the result of convention. At the same time, however, Boyle held that our classificatory practices are subject to constraints. We identify some of these constraints by turning to Boyle's discussion of the late scholastic debate about the plurality of forms, in particular the contributions of Jacopo Zabarella and Daniel Sennert. In this way, we clarify how Boyle strikes a balance between realist and conventionalist elements in his treatment of kinds.