Anxious individuals have been shown to interpret others’ facial expressions negatively. However, whether this negative interpretation bias depends on the modality and familiarity of emotional cues remains largely unknown. We examined whether trait-anxiety affects recognition of multisensory emotional cues (i.e., face and voice), which were expressed by actors from either the same or different cultural background as the participants (i.e., familiar in-group and unfamiliar out-group). The dynamic face and voice cues of the same actors were synchronized, and conveyed either congruent (e.g., happy face and voice) or incongruent emotions (e.g., happy face and angry voice). Participants were to indicate the perceived emotion in one of the cues, while ignoring the other. The results showed that when recognizing emotions of in-group actors, highly anxious individuals, compared with low anxious ones, were more likely to interpret others’ emotions in a negative manner, putting more weight on the to-be-ignored angry cues. This interpretation bias was found regardless of the cue modality. However, when recognizing emotions of out-group actors, low and high anxious individuals showed no difference in the interpretation of emotions irrespective of modality. These results suggest that trait-anxiety affects recognition of emotional expressions in a modality independent, yet cultural familiarity dependent manner.