The goal of this study was to test in a clinical sample the interrater reliability and convergent validity of the Differentiation-Relatedness Scale (D-RS), a measure that evaluates mental representations based on open-ended descriptions of self and significant others. The study also investigated the ability of the D-RS to predict personality disorders (PDs) from Section II of the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5), and the dysfunctional trait domains presented in the Alternative DSM-5 Model for Personality Disorders Criterion B in Section III of the DSM-5. We also evaluated if the D-RS predicts observed Section II PDs over and above Criterion B of the Alternative DSM-5 Model for Personality Disorders. We found that the interrater reliability of the D-RS was good on the basis of the mean scores of 6 independent raters and that it showed moderate convergent validity. Results of dominance analyses indicated that the D-RS is a significant predictor of Section II borderline PD and of the overall number of DSM-5 PDs. When we considered the Section III Criterion B for PDs, the D-RS was not able to predict any of the Personality Inventory for DSM-5 domains, suggesting that the D-RS may be more related to personality functioning behind mental representations than to maladaptive personality traits. Finally, results of hierarchical regression analyses suggested that the D-RS produced a significant but modest increase in the prediction of borderline PD traits and the overall number of PDs traits even when the effect of the Personality Inventory for DSM-5 domains were controlled for.