The methodological risks of defining religion a priori are discussed. The alternative is to study ideas, groups, and institutions that are defined, perceived or contested as religious. In any given case, the dimension of religion that is at stake is identified (specifically, one or all of the three Bs: belief, belonging or behaving) along with which specific interactions between the three are at play for both political and religious communities. The interaction between the two sets of three Bs, religious and political, can explain the politicization of religion and vice versa in specific contexts. From these findings, three trends emerge that cut across all the cases and would be worth further exploration: the loss of local autonomy of religious communities, the securitization of religion and the blurring of the national and international political division. In closing, the respective limits of variable-centered investigations and postcolonial studies are discussed.