This chapter aims to identify the variables influencing central governments that collaborate with secessionist forces to organise an independence referendum. This statistical analysis is based on the PIRA dataset, which comprises 130 cases of independence referendums held between 1944 and 2024. This dataset includes 77 independence referendums held in democratic regimes and 53 cases organised in authoritarian countries. The logistic regression confirms the accuracy of Qvortrup’s competition/proximity model adapted to the acceptance of independence referendums. However, this model must be supplemented with additional variables, such as the level of democracy or territorial contiguity. In sum, while democratic governments have no problem holding self-determination referendums with their former colonies, they are much more reluctant to do so with regions on their mainland. Drawing on these findings, we propose a predictive model of agreed referendums in democratic regimes.