Background: Threat-related attention biases have been implicated in the etiology and maintenance of anxiety disorders. As a result, attention bias modification (ABM) protocols have been employed as treatments for anxious adults. However, they have yet to emerge for children. A randomized, double-blind placebo-controlled trial was conducted to examine the efficacy of an ABM protocol designed to facilitate attention disengagement from threats, thereby reducing anxiety and stress vulnerability in children. Methods: Participants were 34 chronically high-anxious 10-year-olds. An emotional attention spatial cueing task was used. In the ABM condition (n = 18), threat faces never cued the targets' locations, such that the valid-invalid ratio was 0%/100%, respectively. The valid-invalid ratio on neutral cue trials was 25%/75%, respectively. In the control condition, the valid-invalid ratio was 25%/75% for both neutral and threat faces. Anxiety and depression were measured pre- and post-training and pre- and post-stress induction. Results: ABM facilitated attention disengagement from threat. In response to the stressor task, children in the ABM condition reported less state anxiety relative to controls. Conclusion: Computerized attention training procedures may be beneficial for reducing stress vulnerability in anxious children.