The study explores the impact of local ethnic demography on ethnic voting, focusing on an understudied democratic experiment in Southeastern Europe during the 1920s—Yugoslavia. Using original data on 378 counties across four parliamentary elections, the study tests two competing hypotheses regarding inter-ethnic contact between Serbs and Croatians: the threat hypothesis and the contact hypothesis. The findings indicate that in contexts where demographic distribution nears parity between rival ethnic groups, minority nationalist voting intensifies while majority nationalist voting diminishes. Although majority nationalists benefit from the overall ethnic diversity, they lose to liberals in scenarios where there is parity with the rival side of the cleavage conflict.