We explore theoretically and empirically whether social interaction, including local and global interaction, influences the incidence of corruption. We first present an interaction-based model on corruption that predicts that the level of corruption is positively associated with social interaction. Then we empirically verify the theoretical prediction using within-country evidence at the province-level in China during 1998 to 2007. Panel data evidence clearly indicates that social interaction has a statistically significantly positive effect on the corruption rate in China. Our findings, therefore, underscore the relevance of social interaction in understanding corruption.