How do occupying governments ensure their enduring control over territory? One common—though understudied—option has been the plebiscite, an institution that allows individuals residing in conquered territories to decide through a vote whether they want to continue living under the occupier or revert to the pre-occupation status quo. This use of the plebiscite has been common over the past 150 years, having been used (or promised) by German occupiers of Danish Northern Schleswig and Polish/Czech Upper Silesia; Chilean occupiers of Peruvian Tacna; and Russian occupiers in Crimea and Donbas. The use of a referendum presents a clear potential benefit for occupiers: it can provide a popular legitimation for annexation. This matters not only for domestic perceptions of the legitimacy of occupier rule but also for international audiences. Yet, plebiscites also carry a risk: if the occupier loses, it may be deprived of a claim to legitimate control over the occupied territory. Drawing on a novel quantitative analysis of individual-level plebiscite records from Chilean-occupied Tacna in the 1920s, we explore how governments minimize the risk of a failed plebiscite. We highlight the central role of the firm, developing and testing a theoretical framework that reveals a key prediction: occupiers often conduct electoral fraud through private companies.