The word borderline, unlike other clinical descriptive terms, has no obvious affective or action referent in the patient. This observation may shed light on the kind of misalliance so endemic to the work that occurs with these patients. The possibility that the diagnosing of “borderline” might reflect a misrecognition of who has what problem within an analytic treatment, that it might seriously affect the frame of the treatment, and that it might be an act of projective identification by the analyst are explored. I argue that (a) borderline pathology is not an entity, but rather the vast developmental territory of severe personality disturbance, and (b) Winnicott's theoretical contributions to an understanding of the psychopathology of the dyad, particularly around boundary development, are especially helpful.