Childhood maltreatment has been associated with greater attention bias to emotional information, but the findings are controversial. Recently, a novel index of attention bias, i.e., attention bias variability (ABV), has been developed to better capture trauma-related attentional dysfunction. However, ABV in relation to childhood trauma has not been studied. Here, we examined the association of childhood maltreatment history with attention bias/ABV in 128 healthy adult women. Different types of childhood maltreatment were assessed with the Childhood Trauma Questionnaire. Attention bias/ABV was measured by the dot-probe task. Possible mechanisms whereby childhood maltreatment affects attention bias/ABV were also explored, focusing on blood proinflammatory markers and the BDNF Val66Met polymorphism. We observed a significant positive correlation between childhood emotional abuse and ABV (P = 0.002). Serum high-sensitivity tumor necrosis factor-α levels were significantly positively correlated with ABV (P < 0.001), but not with childhood maltreatment. Jonckheere-Terpstra trend test showed a significant tendency toward greater ABV with increasing numbers of the BDNF Met alleles (P = 0.021). A two-way analysis of variance further revealed that the genotype-by-emotional abuse interaction for ABV was significant (P = 0.022); individuals with the Val/Met and Met/Met genotypes exhibited even greater ABV when childhood emotional abuse was present. These results indicate that childhood emotional abuse can have a long-term negative impact on emotional attention control. Increased inflammation may be involved in the mechanism of ABV, possibly independently of childhood maltreatment. The BDNF Met allele may dose-dependently increase ABV by interacting with childhood emotional abuse.