BackgroundDSM-5 may mark the shift from a categorical classification of personality pathology to a dimensional system. Although dimensional and categorical conceptualizations of personality pathology are often viewed as competing, it is possible to develop categories (prototypes) from combinations of dimensions. Robust prototypes could bridge dimensions and categories within a single classification system.MethodTo explore prototype structure and robustness, we used finite mixture modeling to identify empirically derived personality pathology prototypes within a large sample (n=8690) of individuals from four settings (clinical, college, community, and military), assessed using a dimensional measure of normal and abnormal personality traits, the Schedule for Nonadaptive and Adaptive Personality (SNAP). We then examined patterns of convergent and discriminant external validity for prototypes. Finally, we investigated the robustness of the dimensional structure of personality pathology.ResultsThe resulting prototypes were meaningful (externally valid) but non-robust (sample dependent). By contrast, factor analysis revealed that the dimensional structures underlying specific traits were highly robust across samples.ConclusionsWe interpret these results as further evidence of the fundamentally dimensional nature of an empirically based classification of personality pathology.