Anxious individuals have been shown to be hyper-sensitive to cues for others’ negative emotional states. As most studies used facial expressions as emotional cues, we examined whether the trait anxiety affects the cross-modal perception of emotion in the face and voice that were simultaneously presented. The face and voice cues conveyed either matched (e.g., both positive) or mismatched emotions (e.g., positive face and negative voice). Participants indicated the perceived emotion in one of the cues, while ignoring the other. The results showed that highly anxious individuals, compared with low anxious ones, were less able to disregard the to-beignored negative cues and misinterpret the attended happy cues as negative. These results were found regardless of the cue modality. The trait anxiety may affect the integration of emotion in the face and voice.