Few experiences direct affect, behavior, and cognition as thoroughly as feelings of power and powerlessness. The present study examined 403 participants’ narrated experiences feeling powerful ( n = 196) or powerless ( n = 207) using Linguistic Inquiry and Word Count (LIWC) analysis to identify social contexts that might explain the effects of power on emotion expression. Powerful narratives contained more frequent communication about work, whereas powerless narratives contained more frequent communication about sex. Moreover, powerless narratives conveyed greater negative emotionality. A parallel mediation analysis revealed that communication about work and sex helped to explain the association between self-reported feelings of power and expressions of negative emotionality. When participants felt powerful and communicated about work, they expressed lower negative emotionality; when participants felt powerless and communicated about sex, they expressed higher negative emotionality. Modest differences in emotional expression between women and men indicated that power research should report analyses including gender as a control variable. Findings provide direction to the next wave of power research, which should examine organizations and intimate relationships as contexts in which power dynamics are salient.