Emotions shape behavior, but there is some debate over the manner in which they do so. The authors propose that how emotions shape behavior depends, in part, on how people expect emotions to shape behavior. In Study 1, angry (vs. calm) participants made more money in a negotiation when they expected anger to be beneficial. In Study 2, angry (vs. calm) participants killed more enemies in a computer game when they expected anger (but not calmness) to promote performance. In Study 3, excited (vs. calm) participants were more creative when they expected excitement to promote performance, whereas calm (vs. excited) participants were more creative when they expected calmness to promote performance. These findings demonstrate that, at least sometimes, what emotions do depends on what we expect them to do.