In this paper, the electoral geography approach is applied to referendums for the first time. In terms of the methodology, the advantage of aggregated data is that it provides harmonized material for comparing nine referendums held since 1958. The very similar geographical distribution of the "yes" vote in the Gaullist and post-Gaullist referendums (1958–“72), which was identical to the right-wing vote during the second round of presidential elections, ends up supporting the theory suggested by opinion polls on voting motivations that these referendums were "presidentialized". During the past three referendums, the care taken by the president of the republic to distance himself from the referendum seems to have made it possible to limit any potential influence he may personally have had over the result, and this is also manifested in the more specific geographic distribution of the votes.