Functional neuroimaging recently has been used to localize brain dysfunction in borderline personality disorder (BPD). Initial studies have examined baseline activity or emotional reactivity, and our group has investigated what we consider to be a crucial interaction between negative emotion and behavioral (dys)control. This research is beginning to identify abnormal frontolimbic circuitry likely underlying core clinical features of this condition. We review the evidence for dysfunction in specific frontolimbic regions, leading to a mechanistic model of symptom formation in BPD. In addition, we offer an integration of these neuroimaging findings with developmental perspectives on the emergence of borderline psychopathology, focusing on the ways in which early psychosocial experience may interact with developing brain systems. We also consider possible mechanisms of psychotherapeutic change at the neural systems level in BPD. Finally, we propose that future neuroimaging studies of BPD should integrate multiple levels of observation (structural, functional, neurochemical, genetic, and clinical) in a model-driven fashion to further understand the dynamic relationship between biological and psychological factors in the development and treatment of this difficult condition.