I critique the essentialistic notion that psychiatric disorders should be conceptualized as natural kinds, that is, exhaustively defined with reference to inherent properties. Biomedical model thinkers believe that psychiatric natural kinds can best be isolated by studying underlying biopathological processes, while research-oriented clinical psychologists think they can be identified statistically. Both groups assume that if a category can't be conceptualized as a natural kind, it is an arbitrary category. I argue that conceptualizing psychiatric disorders as bounded entities in nature is inconsistent both with medicine's understanding of disease and evolutionary biology's understanding of species. In contrast to natural kinds, I introduce the concept of practical kinds, which are stable patterns that can be identified with varying levels of reliability and validity. I claim that thinking anti-essentialistically and conceptualizing psychiatric disorders as practical kinds is more consistent with a scientific view of the world.