Clinical data suggest that Kernberg’s description of splitting as a defense mechanism is useful in conceptualizing the psychological consequences of abuse in childhood in certain patients. The splitting in these patients is similar to his description of splitting in borderline patients in that it compartmentalizes and sequesters certain overwhelming and painful ego states accompanying negative introjects of the betraying primary object and the betrayed self. These sequestered introjects, furthermore, act as automatons, generating behaviors that arbitrarily re-enact their content even though the patient remains consciously unaware of their historical meaning. Another consequence of the sequestration of these traumatic introjects is that their affects retain their initial power and primitive quality, unmodulated by the usual homogenizing process that is a part of the synthesis of part-object introjects into whole-object introjects; the sequestration, therefore, often painful in itself, must nevertheless be rigidly maintained lest traumatic anxiety in the face of overwhelming affects be re-experienced. Shengold calls the sequence of events that results in this brittle but stubborn painful constriction of the personality “soul murder.” He borrowed the phrase from Freud who used it to refer to what Schreber had suffered at the hands of his sadistic father. That phrase–“soul murder”–may sound melodramatic, but it powerfully conveys what these patients communicate of their experience of themselves. As with Kernberg’s patients, the defensive splitting serves to protect the positive introjects. These patients fear their negative introjects, even more than they feel uncomfortable about the split. They fear their desperate rage will destroy their love objects, and leave them feeling abandoned and hating themselves. As one of my patients put it: “I fear that my destructive anger will leave me all alone in a sea of rubble of my own making.” In the transference, he feared destroying me and our positive bond. In these cases it would seem that the turning to splitting occurred at a later age than it does with Kernberg’s borderline patients. His proposition is that the developmentally normal “splitting,” related to the undifferentiation of the infantile ego, persists as a defensive splitting, perhaps as a consequence of a consistently derailed mother-child dialogue; whereas in my patients it would seem that the normal developmental splitting had waned as ego differentiation proceeded, but that in the face of overwhelming traumata at perhaps 3 or 4 years of age, the primitive defense was invoked regressively.