Kernberg's classification of personality disorders (1987) differentiates psychic organization according to the severity: neurotic, borderline and psychotic. Lenzenweger et al. (2001) used a reduced version of IPO with 57 items developed by Kernberg and Clarkin (1995).IPO was applied in a sample of patients and a control group. We expected to find an adequate reliability and validity of the inventory. Scales adequately distinguish content borderline, neurotic and psychotic.Participants: 288 subjects (64.9% women), 116 patients attended to private clinical practice from February 2007 to September 2009. 172 control subjects matched by sex, social class and sincerity (EPI).Transversal design, a measure collective in the comparison group and individual in patients ones. A group of patients was selected for the retest (n = 88).Instruments. We applied IPO, the BPRS, MCMI-II and MIPS. Diagnoses according to DSM-IV-TR.Internal consistency (Cronbach) was adequate for the three scales: .83; .90 and .89. The testretest reliability was correct for a mean interval of 44 days (.78; .81; .78). The validity analyses differed between diagnostic groups in Axis I (p< .05), but not in the clusters of personality (p>.05). No differences in BPRS with scale of borderline, but yes with neurotic and psychotic ones. The MCMI-II was properly differentiated by the three scales of the IPO.The IPO is an useful scale with reliability and validity. The main drawback concerns certain aspects of the borderline scale.