Attentional capture by emotional faces outside the focus of attention, specifically angry expressions, is often considered to be automatic due to emotional salience. However, previous research has suggested that this capture may depend on factors such as configural face processing, task demands, and attentional competition. This study aimed to examine whether attentional capture by emotional faces is fully automatic or contextdependent using lateralized event-related potentials (ERPs), specifically the N170 and N2pc components, across three experiments. In Experiment 1, emotional faces (angry, happy, and neutral) were paired with houses to assess attentional capture in a non-competitive context. The results showed a significant N2pc response to angry faces, indicating a strong attentional capture. Experiment 2 introduced inverted faces to disrupt configural processing, resulting in an absence of N2pc for inverted faces, suggesting that intact face perception is required for attentional capture by emotional faces. In Experiment 3, two competing faces (angry or happy vs. neutral) were presented, leading to no significant difference in N2pc, indicating that attentional competition neutralized the effect. The N170 component remained unaffected by facial expressions or task demands in Experiments 1 and 2, reflecting its role in structural rather than emotional face processing, and it was absent in Experiment 3. These findings suggest that attentional capture by emotional faces is not entirely automatic and requires intact configural processing and minimal competition. This study challenges the assumption of automatic attention capture by emotional stimuli and emphasizing the importance of both bottom-up and top-down mechanisms.