Limited prospective research has examined whether attention biases to emotion moderate associations between Behavioural Inhibition (BI) and anxiety in preschool-aged children. Furthermore, there has been an over-reliance on behavioral measures in previous studies. Accordingly, we assessed anxiety in a sample of preschool-aged children (3-4 years) at baseline, and again approximately 6 and 11 months later, after they started school. At baseline, children completed an assessment of BI and an EEG task where they were presented with angry, happy, and neutral faces. EEG analyses focused on ERPs (P1, P2, N2) associated with specific stages of attention allocation. Interactions between BI and emotion bias (ERP amplitude for emotional versus neutral faces) were found for N2 and P1. For N2, BI was significantly associated with higher overall anxiety when an angry bias was present. Interestingly for P1, BI was associated with lower overall anxiety when a happy bias was absent. Finally, interactions were found between linear time and happy and angry bias for P1, with a linear a greater decrease in anxiety over time when biases were high. These results suggest that attention to emotional stimuli moderates the BI-anxiety relationship across early development.