Borderline personality disorder is a particularly devastating, yet understudied form of psychopathology. One of the most significant gaps in existing knowledge is the lack of systematic, prospective empirical attention to the developmental precursors of borderline personality. The present investigation was an exploratory attempt to address this limitation through (a) development of a psychometrically sound self-report instrument that assesses borderline personality features in childhood, the Borderline Personality Features Scale for Children (BPFS-C); (b) examination of the stability of BPF in childhood; (c) evaluation of gender differences in BPF in childhood; and (d) evaluation of the specificity of the BPFS-C for assessing borderline personality features. These goals were achieved through the prospective study of a normative sample of 400 (54% female) fourth though sixth graders who were assessed during the Fall of Year 1, Spring of Year 1, and Fall of Year 2. The use of linear mixed modeling techniques provided evidence for the construct validity of the BPFS-C. Further, borderline personality features as assessed with the BPFS-C were found to be moderately stable over the course of the study, with girls reporting higher levels of BPF than boys. Results also demonstrated that children's scores on the BPFS-C were uniquely related to indicators of borderline personality pathology above and beyond their scores on the Children's Depression Inventory. The implications of these results for the study of the development and etiology of borderline pathology are discussed.