Background Attention bias modification treatment (ABMT) targets threat‐related attention biases in anxiety disorders. Most clinical trials of ABMT have focused on adults or small samples of youth. The current randomized controlled trial (RCT) examines ABMT efficacy in youth with social anxiety disorder (SAD) and tests possible moderators of treatment outcomes. Method Sixty‐seven youth with SAD were randomly assigned to ABMT or attention control training (ACT) conditions. Anxiety severity was measured at baseline, posttreatment, and 3‐month follow‐up. ClinicalTrials.gov name and identifier: Attention bias modification treatment for children with social anxiety, NCT01397032; http://www.clinicaltrials.gov. Results Both ABMT and ACT induced significant reductions in clinician and self‐rated social anxiety (ps < .001). An additional reduction was observed at the 3‐month follow‐up in clinician‐rated anxiety symptoms (p = .03). Moderation effects were nonsignificant for the clinician‐rated anxiety outcome, but age moderated self‐reported anxiety. Older but not younger children, showed significant reduction in anxiety following ABMT relative to ACT (p < .001). Individual differences in attention control also moderated ABMT's effect on self‐reported anxiety (p = .05). Children rated by their parents as lower on attention control benefited more from ABMT than those rated higher on attention control. Baseline attention bias did not moderate anxiety (p = .17). Conclusions Despite significant reductions in social anxiety, no specific evidence for ABMT was found relative to a control condition. Age and attention control moderated ABMT effects on self‐reported SAD symptoms, with clinical effects for older relative to younger children and for those with lower attention control. These results highlight the need to consider developmental influences in the implementation of ABMT protocols.