The object of this study is to assess the internal validity of DSM-III personality constructs and to explore whether the constituent elements are better explained by an alternate internally coherent classification. A two-stage stratified random sample of subjects identified at the Baltimore site of the Epidemiologic Catchment Area (ECA) program were examined by psychiatrists for DSM-III personality attributes using a semistructured instrument. Dichotomous factor analysis was used in the confirmatory mode to test whether a single factor explained each of the 11 DSM-III personality disorders. This approach rejected a single explanatory factor for all but compulsive personality disorder. Exploratory factor analysis showed that these DSM-III personality features are parsimoniously described by a five-factor model. These factors are warmth, animation, timidity, trust, and scrupulousness.