Research on the determinants of election fraud typically focuses on conventional factors such as election monitoring, legal punishments, and voter education. This paper examines an often-overlooked factor: whether polling location (e.g., school, place of worship) influences election fraud. Combining validated fraud measures from the 2009 Afghan presidential election with a novel instrumental variable approach, we find that polling centers within schools report an 8 percentage-point lower likelihood of fraud compared to those within mosques. Two mechanisms may explain this difference. First, mosques designated as polling centers were more likely to be attacked by the Taliban, likely suppressing turnout and creating incentives for fraudulent votes through a vote-substitution channel. Second, election-related complaint data indicate systematic differences in the behavior of voters and polling officials between schools and mosques. Compared to mosques, voters at schools are more willing to file complaints, complainants are more often women, and polling officials are less likely to be the subject of complaints.