The current experiment tested whether changing one‘s nonverbal behavior prior to a high-stakes social evaluation could improve performance in the evaluated task. Participants adopted expansive, open (high-power) poses, or contractive, closed (low-power) poses, and then prepared and delivered a speech to two evaluators as part of a mock job interview, a prototypical social evaluation. All speeches were videotaped and coded for overall performance and hireability, and the potential mediators of speech quality (e.g., content, structure) and presentation quality (e.g., captivating, confident). As predicted, high power posers performed better and were more likely to be chosen for hire, and this relationship was mediated only by presentation quality, not speech quality. Power pose condition had no effect on body posture during the social evaluation, thus highlighting the relationship between preparatory nonverbal behavior and subsequent performance.