The aim of this study was to examine patterns of comorbidity among personality disorders (PDs) in a sample of 156 psychiatric inpatients. PDs were assessed with Semistructured Clinical Interview for DSM-III-R Personality Disorders. To determine significant co-occurrence among axis II diagnoses, odds ratio and the percent of co-occurrence of pairs of disorders were calculated. Both statistical methods revealed high rates of comorbidity: significance association was found for 36 pairs of disorders using the percent of co-occurrence, and for 22 pairs of disorders using the odds ratio. These results support the concept of 'apparent comorbidity' for most PDs, deriving from conceptual and definitional artifacts or from a 'state-biasing effect'. In light of these observations, a categorical approach to PDs, resulting in a list of diagnoses, appears useless in psychiatric practice. A dimensional classification is probably better suited for PDs, improving the understanding of personality psychopathology and its clinical implications.