How does the disruption of democratic institutions affect voters’ behaviour once democracy is restored? I answer this question by looking at the uneven repression of elected town councils in Vichy. Due to limited capacity, the Pétain government could not consolidate its territorial control by sending loyal administrators to each municipality. Only towns with a population above 2,000 inhabitants had their mayors replaced. I thus exploit an RDD to highlight the effects of this heterogeneity in democratic disruption onto voters’ behaviour after the restoration of democracy. By looking at the results of the 1945 legislative elections, I find that, in treated municipalities, the MRP – the main non-leftist resistance party – obtained an average 4 p.p. increase in its vote share vis-à-vis municipalities below the cut-off. I find a corresponding contraction in the support for the Radical Party and the Alliance Démocratique, parties not directly linked to the resistance. Interestingly, I find no effect for the Communist and Socialist Parties. This is – I argue – because left-wing parties’ officials (mayors included) were purged regardless of the population cut-off, which explains the absence of significant local effects. Insofar as the mechanisms, I show how a greater degree of democratic repression likely triggered more fierce resistance. In turn, such more intense resistance experience translated into more support for resistance parties, and preferences for resistance parties materialised into electoral support for the ideologically closest resistance party, that is, in relatively less left-wing municipalities, the MRP.