In this investigation, gene–environment–gender interaction effects in predicting child borderline personality disorder symptomatology among maltreated and nonmaltreated low-income children (N = 1,051) were examined. In the context of a summer research camp, adult-, peer-, and self-report assessments of borderline precursor indicators were obtained, as well as child self-report on the Borderline Personality Features Scale for Children. Genetic variants of the oxytocin receptor genotype and the FK506 binding protein 5 gene CATT haplotype were investigated. Children who self-reported high levels of borderline personality symptomatology were differentiated by adults, peers, and additional self-report on indicators of emotional instability, conflictual relationships with peers and adults, preoccupied attachment, and indicators of self-harm and suicidal ideation. Maltreated children also were more likely to evince many of these difficulties relative to nonmaltreated children. A series of analyses of covariance, controlling for age and ancestrally informative markers, indicated significant Maltreatment × Gene × Gender three-way interactions. Consideration of the maltreatment parameters of subtype, onset, and recency expanded understanding of variation among maltreated children. The three-way interaction effects demonstrated differential patterns among girls and boys. Among girls, the gene–environment interaction was more consistent with a diathesis-stress model, whereas among boys a differential-sensitivity interaction effect was indicated. Moreover, the genetic variants associated with greater risk for higher borderline symptomatology, dependent on maltreatment experiences, were opposite in girls compared to boys. The findings have important implications for understanding variability in early predictors of borderline personality pathology.