Using individual-level data provided by the Demographic and Health Survey (2000, 2004, 2010) merged by GPS-based coordinates with community-level data sourced from the Integrated Household Survey (2010-2011) in Malawi, this paper asks the question of whether religious teachings affect women’s marital practices. To address endogeneity associated with an individual’s religious affiliation, the analysis takes an instrumental variable (IV) approach by exploiting a unique setting of the Christian mission, which dates back to the late 19th century. Being exposed to the mission, measured by geographical distance (km) to the influential station, Livingstonia, enabled the indigenous population to gradually convert to Christianity when they were not the Yao, an ethnic group that was proselytized into Islam because of their ivory and slave trades with the Arabs that had existed prior to the arrival of the mission. The IV approach shows that the religious affiliation cannot always be taken as exogenous in an empirical analysis. Using the distance-ethnicity (non-Yao) interaction as an IV for the religious distribution of the current generation with a control of ethnicity-level historical covariates (e.g., slave exports, access to railway networks) and abundance of community-level local conditions (e.g., geography, climate) or community-fixed effects, this study finds that compared to those practicing the other religions (Islam and other) or no religion, Christian females are more likely to postpone their marriage by 1-3 years with less likelihood of engaging in polygynous relationship by approximately 30 percentage points. Formal educational attainment, fear of HIV infection, and religion-based segmentation in both the marriage and labor markets are less likely to explain the religious effects. While it is difficult to completely rule out alternative mechanisms, these findings are consistent with the view that religious education plays a role.