Context: Attentional bias modification (ABM) procedures is a promising intervention tool for a variety of clinical conditions. Objectives: This study provides an updated review of the clinical impact of ABM by employing standard meta-analytic procedures to: (a) estimate the average effect size of ABM in reducing both attention bias (AB) and symptoms; (b) estimate the average effect size for different conditions (e.g., anxiety, depression, substance abuse); (c) test possible variables that may moderate the effect sizes, and (d) investigate the relationship between pre-existent AB and the reduction in AB and symptoms. Method: We included 43 controlled trials with a total of 2268 participants providing 47 group comparisons (i.e., training vs. control condition). Inclusion criteria were: AB was specifically targeted to reduce symptomatology and/or emotional vulnerability; participants were randomized to the experimental conditions; a control condition (defined as sham training) existed; symptoms were assessed at least post-intervention; sufficient data were provided to allow effect size estimation. Results: We obtained a small overall effect size on symptoms post intervention, g = 0.160, 95% CI = [0.055; 0.265], driven by anxiety studies, g = 0.260, 95% CI = [0.132; 0.388], and studies conducted in healthy participants, g = 0.211, 95% CI = [0.046; 0.375]; no significant effect sizes were found post intervention for other symptom categories. Conclusion: The therapeutic benefit of ABM is rather small for anxiety, while the amount of data for other symptom categories is limited. We argue that more efficient, psychometrically sound procedures are needed for assessing and modifying AB.