Existing research shows that appeals to self-interest sometimes increase and sometimes decrease prosocial behavior. We propose that this inconsistency is in part due to the framings of these appeals. Different framings generate different salient reference points, leading to different assessments of the appeal. Study 1 demonstrates that buying an item with the proceeds going to charity evokes a different set of alternative behaviors than donating and receiving an item in return. Studies 2 and 3a-g establish that people are more willing to act, and give more when they do, when reading the former framing than the latter. Study 4 establishes ecological validity by replicating the effect in a field experiment assessing participants’ actual charitable contributions. Finally, Study 5 provides additional process evidence via moderation for the proposed mechanism. We discuss theoretical and practical implications of these findings and suggest avenues for future research.