The author reviews the DSM-III section on personality disorders, discusses several of its more controversial diagnoses, and suggests some possible alternatives. He attributes the continued low reliability of personality diagnoses, compared with the other major sections of DSM-III, to two inherent obstacles: the lack of clear boundaries demarcating the personality disorders from normality and from one another, and the confounding influence of state and role factors. Nonetheless, the DSM-III multiaxial system highlights the importance of personality diagnosis and, together with the provision of clearly specified diagnostic criteria, achieves a considerably improved reliability compared with previous nomenclatures.