Interest in participatory forms of constitution making continues to grow and with it, interest in the classic means of citizen involvement: the referendum. Many scholars can think of specific examples of a referendum that went notably well or poorly. However, we have to this point lacked a comprehensive historical account of when, how, and to what effect referenda have been used in attempts to ratify constitutional reforms. In this chapter, we introduce a new dataset that includes information about every identifiable constitutional referendum between 1789 and 2016 (n = 644). Some of our findings accord with casual impressions: referenda are increasingly common both as explicit provisions in constitutions and as events, and are much more likely to pass in authoritarian settings than they are in consolidated democracies. One startling finding from these records is that referenda on new (replaced) constitutions are much more likely to pass than are those on amendments. We consider this discrepancy in terms of both elite and popular decision-making. On the elite side, we suggest that leaders are likely to have more control of the process with constitutional replacement than with revision. We also provide some preliminary arguments about how differences in the evaluation of risk may drive voter behavior.