The DSM-IV section of the DSM-IV and ICD-10 Personality Questionnaire (DIP-Q) was used to screen for personality disorders in 448 subjects from three clinical samples (general and forensic psychiatric patients and candidates for psychotherapy) and a sample of 139 healthy volunteers. Differences between the samples with regard to patterns of personality pathology in relation to concurrent Axis I disorders and sociodemographic variables were analysed. The prevalence of personality disorders according to DIP-Q was 14% among the healthy volunteers, compared to 59% in the general psychiatric sample, 68% in the forensic psychiatric sample and up to 90% among psychotherapy candidates. Moreover, from a dimensional perspective (i.e. the number of fulfilled Axis II criteria), all clinical groups differed significantly from the control group in all specified personality dimensions and clusters. Dimensional DIP-Q cluster scores also discriminated significantly between the three clinical samples. Unexpectedly, the odds ratio for an Axis II disorder was nearly five times higher among psychotherapy applicants than among general psychiatric patients, independent of concomitant Axis I disorders, gender or age. The strongest association between DIP-Q score and Axis I disorders was found for depressive disorders, which more than doubled the odds ratio for a personality disorder diagnosis. This association could result from high true comorbidity, but could also be due to the fact that a concomitant depressive state can increase self-reported personality difficulties. The high prevalence among psychotherapy candidates may to some extent reflect help-seeking exaggeration of problems. These are aspects to consider when using the DIP-Q, which overall appears to discriminate well between different samples.