Voting is perceived as free and rational. Citizens make whatever choices they wish, shielded from external influences by the privacy of the voting booth. The current paper, however, suggests that a subtle source of influence—polling places themselves—can impact voting behavior. In two elections, people voting in churches were more likely to support a conservative candidate and a ban on same-sex marriage, but not the restriction of eminent domain. A field experiment found that people completing questionnaires in a chapel awarded less money (relative to people in a secular building) to insurance claimants seeking compensation for abortion pills, but not to worker's compensation claimants. A laboratory experiment found that people subliminally exposed to ecclesiastical images awarded less money (relative to people exposed to control images) to abortion pill claimants, but not to worker's compensation claimants. Exposure to ecclesiastical images affected only Christians; non-Christians' awards were unaffected by the prime. These findings show that polling locations can exert a powerful and precise influence on political attitudes and decision making.