This study examines the inter-rater agreement of comorbid DSM-IV personality disorders in substance abusers. Researchers evaluated 75 patients with substance use disorders using DSM-IV criteria and rating scales to assess the severity of each disorder. The results showed moderate convergent validity for cluster B disorders and moderate discriminant validity for eight of the ten personality disorders. However, convergent validity for antisocial and narcissistic personality disorders was only moderate. The study found that dimensional ratings may be used in research and clinical practice with caution, and may be collected as one of several sources of information to describe the personality of a patient. Clinicians rated patients based on their knowledge of the patients, and the results showed that convergent and discriminant correlations were mostly as good as those based on interviews or questionnaires. The study found that the reliability of staff observations of 8 out of 10 disorders was supported at the given level of alpha (0.005). However, schizoid and avoidant personality disorders were exceptions, as they are related to introversion and interpersonal withdrawal. The use of dimensional scores increased the power of the study substantially. The study found that inter-rater agreement on personality disorders was better for criterion-based ratings than for simple rating scales. However, the difference was not large. The correlation of raters' ratings of personality disorders with another rater in this sample ranged from 0.29 for schizoid personality to a maximum of 0.67 for narcissistic personality disorder. While convergent validity was improved compared with rating scales, the amount of high discriminant correlations was still substantial. The study also found that the use of dimensional scores may not be sufficient for certain research purposes, especially for schizoid and avoidant traits. The study had some limitations, including a heterogeneous sample group and the absence of self-reported data on personality disorders. Despite these limitations, the study supports the reliability of clinicians' ratings of personality disorders, with the exception of schizoid and avoidant personality disorders.This study examines the inter-rater agreement of comorbid DSM-IV personality disorders in substance abusers. Researchers evaluated 75 patients with substance use disorders using DSM-IV criteria and rating scales to assess the severity of each disorder. The results showed moderate convergent validity for cluster B disorders and moderate discriminant validity for eight of the ten personality disorders. However, convergent validity for antisocial and narcissistic personality disorders was only moderate. The study found that dimensional ratings may be used in research and clinical practice with caution, and may be collected as one of several sources of information to describe the personality of a patient. Clinicians rated patients based on their knowledge of the patients, and the results showed that convergent and discriminant correlations were mostly as good as those based on interviews or questionnaires. The study found that the reliability of staff observations of 8 out of 10 disorders was supported at the given level of alpha (0.005). However, schizoid and avoidant personality disorders were exceptions, as they are related to introversion and interpersonal withdrawal. The use of dimensional scores increased the power of the study substantially. The study found that inter-rater agreement on personality disorders was better for criterion-based ratings than for simple rating scales. However, the difference was not large. The correlation of raters' ratings of personality disorders with another rater in this sample ranged from 0.29 for schizoid personality to a maximum of 0.67 for narcissistic personality disorder. While convergent validity was improved compared with rating scales, the amount of high discriminant correlations was still substantial. The study also found that the use of dimensional scores may not be sufficient for certain research purposes, especially for schizoid and avoidant traits. The study had some limitations, including a heterogeneous sample group and the absence of self-reported data on personality disorders. Despite these limitations, the study supports the reliability of clinicians' ratings of personality disorders, with the exception of schizoid and avoidant personality disorders.