Background: The assessment of personality disorder is currently inaccurate, largely unreliable, frequently wrong and in need of improvement.
Aims: To describe the errors inherent in the current systems and to indicate recent ways of improving personality assessment.
Method: Historical review, description of recent developments, including temporal stability, and of studies using document-derived assessment.
Results: Studies of interrater agreement and accuracy of diagnosis in complex patients with independently established personality status using document-derived assessment (PAS-DOC) with a four personality cluster classification, showed very good agreement between raters for the flamboyant cluster B group of personalities, generally good agreement for the anxious/dependent cluster C group and inhibited (obsessional) cluster D group, but only fair agreement for the withdrawn cluster A group. Overall diagnostic accuracy was 71%.
Conclusions: Personality function or diathesis, a fluctuating state, is a better description than personality disorder. The best form of assessment is one that uses longitudinal repeated measures using a four-dimensional system.