Over the past twenty years, several taxonomies of personality and psychopathology have been developed. More recently, many studies have compared dimensional models of personality pathology to categorical diagnoses of personality disorders. Altogether, this proliferation of research suggests the value of articulating the desirable properties of a good taxonomic system. Here, the authors extend basic research in cognitive science on the limitations of representational capacity, which suggests that humans need to compress complex clinical presentations to make good judgments. With this in mind, the authors propose that information compression and information fidelity are two principles that are essential to good taxonomy. The principle of information compression is that taxonomies should prune the complexities of a detailed clinical presentation to focus on important sources of covariation. The principle of information fidelity is that a good taxonomy should maintain essential features that reasonably approximate the structure of an individual or the population. They conclude with the claim that the overarching goal of taxonomic science in classifying personality pathology is to provide clinicians and researchers with empirically based informative priors that help to bias thinking toward useful clinical distinctions.