This dissertation examines psychiatry from a philosophy of science perspective, focusing on issues of realism and classification. Questions addressed in the dissertation include: What evidence is there for the reality of mental disorders? Are any mental disorders natural kinds? When are disease explanations of abnormality warranted? How should mental disorders be classified? -/- In addressing issues concerning the reality of mental disorders, I draw on the accounts of realism defended by Ian Hacking and William Wimsatt, arguing that biological research on (...) mental disorders supports the inference that some mental disorders (e.g., schizophrenia, mood disorders, and anxiety disorders) are real theoretical entities, and that the evidence supporting this inference is causal and abductive. In explicating the nature of such entities, I argue that real mental disorders are natural kinds insofar as they are natural classes of abnormal behavior whose members share the same causal structure. I present this position in terms of Richard Boyd’s homeostatic cluster property theory of natural kinds, and argue that this perspective reveals limitations of Hacking’s account on the looping effects of human kinds, which suggests that the objects classified by psychiatrists are unstable entities. I subsequently argue that a subset of mental disorders (e.g., schizophrenia and Down syndrome) are mental illnesses insofar as they are disorders caused by a dysfunctional biological process that leads to harmful consequences for individuals. I present this analysis against Thomas Szasz’s argument that mental illness is a myth. -/- In addressing issues of psychiatric classification, my analysis focuses on the Diagnostic and Statistical Manual of Mental Disorders (DSM), which has been published regularly by the American Psychiatric Association since 1952, and is currently in its fourth edition. After examining the history of DSM in the twentieth century, and in particular, DSM’s shift to an atheoretical and purely descriptive system in the 1980s, I consider the relative merits of descriptive versus causal systems of classification. Drawing on Carl Hempel’s analysis of taxonomic systems in psychiatry, I argue that a causal classification system would provide a superior approach to psychiatric classification than the descriptive system currently favored by DSM. (shrink)