Objective: Attention Bias Modification Training (ABMT) for anxiety aims to train attention away from threatening stimuli and toward neutral stimuli. Although ABMT shows promising anxiety reduction effects in children and adolescents, no study has examined its influence on neural indicators of attention measured using event-related potentials (ERPs) in children or adolescents (i.e., youths). The present study examined the influence of ABMT on the P1, N170, P2 and P3 ERP components during completion of the emotional faces dot probe task in youths with anxiety disorders who failed to respond to cognitive behavioral therapy. Method: Thirty youths (M age = 11.97, SD = 2.89) with primary DSM-IV-TR anxiety disorders completed the dot probe task while undergoing electroencephalogram (EEG) to obtain ERPs before, immediately after, and eight weeks after eight sessions of either ABMT (n = 14) or a control task regimen (CT), (n = 16). Results: At post-treatment, statistically significant effects were found for P1 and P3 mean amplitudes: P1 was significantly higher during trials showing neutral-neutral (NN) face pairs in the ABMT arm than in the CT arm; P3 was significantly higher during trials showing NN face pairs than during trials showing neutral-threat (NT) face pairs in the ABMT arm, but not the CT arm. At eight-week follow-up, participants in both arms showed significantly higher (more negative) N170 responses for NN trials than for NT trials. Conclusions: Attention Bias Modification Treatment led to increases in neural processing of neutral stimuli in early and late stage attentional processing, as measured by the P1 and P3 components, respectively. These components during the dot probe task are promising neural markers of ABMT’s effects on attentional processing in youth with anxiety disorders.