Researchers have not reached a consensus regarding the effects of religion on crime reduction. In this paper, we revisit the problem by investigating Dajia Matsu pilgrimage, the most preeminent religious festival with different starting dates every year in Taiwan, as a source of variation in religion strength in order to estimate the effect of religion on crimes. Using daily data from 2001 to 2014, we find that, compared to counties with lower shares of residents believing in folk religion, the festival reduced property crime rates more in counties with higher shares. However, in these counties, little evidence is found for a reduction in violent crimes on the days during the procession period. These suggest that the religion does reduce crime, but mainly those of relatively minor property crimes.