Current knowledge of the associations between personality disorders (PDs) and the five-factor model (FFM) is based largely on the results of linear correlation statistics. Yet we do not know whether FFM–PD associations are indeed linear, and correlational statistics are not directly informative regarding the FFM deviations of individuals with PDs. In this study, graphical analyses of FFM–PD associations for a large, clinical and nonclinical combined sample revealed a diversity of linear and nonlinear FFM patterns, at both the domain and facet levels, for most PDs. However, the FFM deviations from normative levels were only moderate. The discussion focuses on the strengths and weaknesses of the FFM approach to PDs. Copyright © 2005 John Wiley & Sons, Ltd.