The paper reports three experiments that investigate the influence of facial cue valence on attentional effects in a cueing paradigm. The studies suggest that trait anxiety is associated with increased attentional dwell-time on emotional facial stimuli, particularly those depicting anger or happiness, relative to neutral faces. Experiment 1 found that participants took longer to respond to targets on invalid trials when the cue was an angry or happy face compared to a neutral face. Experiment 2 used the inhibition of return (IOR) paradigm, increasing the cue-target onset asynchrony to 960 ms, and found that high trait anxious participants showed reduced IOR effects for angry and happy faces, indicating delayed disengagement from these stimuli. Experiment 3 further tested the hypothesis by inducing state anxiety in participants and found that high trait anxious individuals showed reduced IOR for angry faces, but not for jumbled faces, suggesting that the threat signal in angry faces may be particularly salient. These results support the idea that attentional bias in anxiety involves difficulty in disengaging from threat-related and emotional stimuli.The paper reports three experiments that investigate the influence of facial cue valence on attentional effects in a cueing paradigm. The studies suggest that trait anxiety is associated with increased attentional dwell-time on emotional facial stimuli, particularly those depicting anger or happiness, relative to neutral faces. Experiment 1 found that participants took longer to respond to targets on invalid trials when the cue was an angry or happy face compared to a neutral face. Experiment 2 used the inhibition of return (IOR) paradigm, increasing the cue-target onset asynchrony to 960 ms, and found that high trait anxious participants showed reduced IOR effects for angry and happy faces, indicating delayed disengagement from these stimuli. Experiment 3 further tested the hypothesis by inducing state anxiety in participants and found that high trait anxious individuals showed reduced IOR for angry faces, but not for jumbled faces, suggesting that the threat signal in angry faces may be particularly salient. These results support the idea that attentional bias in anxiety involves difficulty in disengaging from threat-related and emotional stimuli.