In this study Kernbergs structural interview, which is directed at the differential diagnosis of neurotic, borderline and psychotic personality structure, is investigated in its interrater reliability. There are two categories psychiatric patients (hospitalized and non-hospitalized) 37 in total; men (n = 20) and women (n = 17); two measuring-instruments (in vivo and video); three judges who change roles. The results indicate that it is possible to attribute these patients to the conditions: neurosis, borderline and psychosis. The Kappa values for interrater reliability vary from moderate to substantial. Further research is meaningful and a well established training can support the use of the structural interview in clinical practice.