A seminal case study by Festinger found, paradoxically, that evidence that disconfirmed religious beliefs increased individuals' tendency to proselytize to others. Although this finding is renowned, surprisingly, it has never been subjected to experimental scrutiny and is open to multiple interpretations. We examined a general form of the question first posed by Festinger, namely, how does shaken confidence influence advocacy? Across three experiments, people whose confidence in closely held beliefs was undermined engaged in more advocacy of their beliefs (as measured by both advocacy effort and intention to advocate) than did people whose confidence was not undermined. The effect was attenuated when individuals affirmed their beliefs, and was moderated by both importance of the belief and open-mindedness of a message recipient. These findings not only have implications for the results of Festinger's seminal study, but also offer new insights into people's motives for advocating their beliefs.