The range of putatively medical animal practices varies widely both functionally and mechanistically. In this article, we argue that the definitions of medicine available in the empirical literature are inadequate for distinguishing genuinely medical practices from other adaptive behaviors. We aim to improve this conceptual landscape by proposing a definition that incorporates both cognitive and functional requirements, enabling finer-grained distinctions across species and behaviors. We apply our definition to the evidence and determine which animal behaviors show a mere difference of degree with paradigmatic medical practices—and should be seen as medicine—and which should be excluded from this nomenclature.