Background. The categorical classification system for personality disorder (PD) has been frequently criticized and several alternative dimensional models have been proposed.

Method. Antecedent, concurrent and predictive markers of construct validity were examined for three models of PDs: the Five-Factor Model (FFM), the Schedule for Nonadaptive and Adaptive Personality (SNAP) model and the DSM-IV in the Collaborative Study of Personality Disorders (CLPS) sample.

Results. All models showed substantial validity across a variety of marker variables over time. Dimensional models (including dimensionalized DSM-IV) consistently outperformed the conventional categorical diagnosis in predicting external variables, such as subsequent suicidal gestures and hospitalizations. FFM facets failed to improve upon the validity of higher-order factors upon cross-validation. Data demonstrated the importance of both stable trait and dynamic psychopathological influences in predicting external criteria over time.

Conclusions. The results support a dimensional representation of PDs that assesses both stable traits and dynamic processes.