Some people claim that religious schools are poorly suited to prepare children for citizenship in a multi-religious society that is (or aspires to be) a liberal democracy. In what sense(s), by what mechanism(s), and to what extent might this be so? And what could be the implications for public policy? I propose an analytical and evaluative framework for addressing these questions. There are several potentially independent dimensions on which a school may have a religious character, and each of these dimensions is a continuous variable. Schools that are strongly religious on all of these dimensions are indeed very poor instruments of civic education in a multi-religious society. But what about schools whose religious character is far weaker on each dimension? If these schools are inferior to their secular counterparts for civic educational purposes, that inferiority may be very slight. Given the great diversity among religious schools, and if--as I argue--the civic goals of education are not the only important values that ought to guide public education policy, there are powerful reasons to discriminate among (proposed) religious schools when making policy decisions about regulation and funding. Those who oppose such a discriminating approach must demonstrate that the benefits of 'difference blindness' in this domain outweigh its substantial costs.