Background: Twin studies in children reveal that familial aggregation of anxiety disorders is due to both genetic and environmental factors. Cognitive biases for threat information are considered a robust characteristic of childhood anxiety. However, little is known regarding the underlying aetiology of such biases and their role in anxiety disorders. Method: A face version of the dot-probe task measuring attentional biases for negative (anger, fear, sad, disgust) and positive (happy) facial expressions was completed by 600, 8-year-old twins; the largest study of its kind. Twin correlations for attentional bias scores were compared to estimate genetic and environmental effects. Parent-report diagnostic interviews identified children with an anxiety disorder. Indices of inferred genetic and familial risk for anxiety disorders were created for each child. Data were analysed using a series of logistic regressions. Results: Anxious children showed greater attentional avoidance of negative faces than nonanxious children; "t" (548) = 2.55, p less than 0.05. Attentional avoidance was not under genetic or shared environmental influence. Risk for anxiety disorders was predicted by familial factors. Both attentional avoidance and inferred familial risk were significant but independent predictors of anxiety disorders (ORs = 0.65 and 3.64, respectively). Conclusions: Anxiety-related attentional biases and familial risk play important but independent roles in childhood anxiety disorders. If replicated, these findings indicate that links between genetic risk and anxiety disorders lie outside the domain of attentional processes. (Contains 3 tables and 1 figure.)