It is pointed out that many published empirical studies on the issue of borderline conditions suffer from methodological shortcomings due to the prevailing empiricist simplification of research. Attention is drawn to the fact that the current psychiatric research in classification issues is dominated by an exaggerated quest for reliability at the expense of concern with validity issues. Concepts of operational criteria, polythetic-prototypic systems, and epistemic peculiarities of psychiatric, clinical typification are briefly exposed. It is suggested that future scientific progress in the area of personality research must take into account these general methodological considerations: repotentiate clinical-phenomenological observation, include prototypical approaches, extend studied samples with subjects drawn from the general population and import insights from personality trait research in normal psychology. Some of the claims of Gunderson and Torgersen (this volume) are discussed.