This article asks why some ethnically distinct regions fight fiercely to secede while others struggle to save the same multinational state. It tests competing explanations using a new dataset containing forty-five cases, significantly more than any previous study in the Soviet setting. The empirical results confirm arguments that the most separatist regions tend to be those possessing the most wealth, containing the least assimilated ethnic groups and already enjoying the greatest levels of autonomy. Demonstration effects are also found to be powerful. No support is found for prominent theories pointing to group upward mobility and ‘skill sets’ as being decisive. Group histories of grievous exploitation or national independence are found not to explain patterns of secessionism.