Emotion perception is not only influenced by the ethnic group-membership of interaction partners but also depends on whether a person is engaged with the encoder of an emotion. To characterize the neural correlates of the influence of ethnicity and engagement on emotion perception, German participants rated the valence of video-sequences while undergoing fMRI. In these video-sequences Asian-looking and European-looking virtual characters expressed a positive (happiness) and a negative (anger) emotion while either gazing directly at the participants or at another agent, thereby varying the perception of engagement. Results show that the ventromedial prefrontal cortex is involved when participants observe ethnic in-group members compared to ethnic out-group members express a positive emotion at them compared to at another person. In contrast, the dorsolateral and dorsomedial prefrontal cortex are involved when participants observe ethnic out-group members compared to ethnic in-group members express a negative emotion at them compared to at another person.