© 2018 Taylor & Francis. The use of personality disorder (PD) categories persists, despite the evidence against them. An often overlooked reason for this is the fact that the true structure underlying the Diagnostic and Statistical Manual of Mental Disorders (DSM) taxonomy is still unknown: We cannot be certain which disorders are valid, and which ones are arbitrary mixtures of heterogeneous traits. To address this gap, we factor analyzed the Personality Diagnostic Questionnaire (PDQ–4+; Hyler, 1994) at the criterion level in a mixed clinical and nonclinical sample of 2,519 individuals. The resulting structure was more similar to current dimensional taxonomies than to the DSM classification at all hierarchical levels. Whereas paranoid and antisocial PDs—and to a lesser extent avoidant, dependent, depressive, and schizoid PDs—were fairly homogeneous, all other disorders turned out to be combinations of 2 or 3 unrelated dimensions. Our results strongly support the structure of empirically based dimensional taxonomies and relocate DSM criteria within this emerging framework, thus contributing to preserving much of the knowledge accumulated to date.