This essay, which seeks to provide an historical framework for our efforts to develop a scientific psychiatric nosology, begins by reviewing the classificatory approaches that arose in the early history of biological taxonomy. Initial attempts at species definition used top-down approaches advocated by experts and based on a few essential features of the organism chosen a priori. This approach was subsequently rejected on both conceptual and practical grounds and replaced by bottom-up approaches making use of a much wider array of features. Multiple parallels exist between the beginnings of biological taxonomy and psychiatric nosology. Like biological taxonomy, psychiatric nosology largely began with 'expert' classifications, typically influenced by a few essential features, articulated by one or more great 19th-century diagnosticians. Like biology, psychiatry is struggling toward more soundly based bottom-up approaches using diverse illness characteristics. The underemphasized historically contingent nature of our current psychiatric classification is illustrated by recounting the history of how 'Schneiderian' symptoms of schizophrenia entered into DSM-III. Given these historical contingencies, it is vital that our psychiatric nosologic enterprise be cumulative. This can be best achieved through a process of epistemic iteration. If we can develop a stable consensus in our theoretical orientation toward psychiatric illness, we can apply this approach, which has one crucial virtue. Regardless of the starting point, if each iteration (or revision) improves the performance of the nosology, the eventual success of the nosologic process, to optimally reflect the complex reality of psychiatric illness, is assured.