How do insurgencies decide their repertoire of violence? Why do some rebel groups recur more to terrorist attacks than to guerrilla-like actions? I propose in this paper a systematic exploration of the repertoire of violence practiced by all insurgent groups in the world from 1970 to 1997. I claim that the repertoire of violence insurgents can use is determined by its capacity to seize and hold territory from the state’s grip. Whenever insurgents are able to control territory, they are better equipped to proceed with the typical guerrilla war against the state. To the contrary, if insurgents do not liberate territory from the state’s hands, they must remain underground, and therefore they cannot but rely on attacks most people would identify as fully terrorist. I use the Global Terrorism Database to test our theoretical claim. Although GTD1 claims to collect only terrorist events in the world from 1970 to 1997, its definitional criterion is so loose that many guerrilla actions were also included, under the rubric of “facility” attacks. I find in the cross-sectional analysis that holding territorial control is a major causal factor of the repertoire of insurgent violence, absorbing the effect of state capacity. In addition to territorial control, the size of the group (recruitment) has also an impact on the types of tactics insurgent groups follow, with larger groups recurring more to facility attacks. I complement this with the study of a specific insurgency, Hezbollah. Hezbollah’s switch in tactics from the initial reliance on bombs to the later adoption of guerrilla tactics was largely anticipated by its capacity to capture and hold territory in the South of Lebanon. Despite seizing territory, Hezbollah still kept planting bombs in the localities remaining beyond its control, as epitomized by the recourse to terrorist attacks abroad.