Studies explaining variation in conflict severity, mostly operationalized as the number battle-related deaths but also human rights violations, focus on characteristics on the monthly or annual level and over longer periods of time looking at first-level administrative regions and countries. They identify factors such as economic conditions, natural resources, and rebel and government characteristics which affect conflict severity. However, these measures do not take civil war dynamics such as changes in territorial control into account and therefore provide an incomplete picture. Territorial control affects rebel tactics used against the government and local civilians. I argue that momentum, specifically rebel or government territorial advances, affects the timing, location and intensity of state-based and one-sided violence. Drawing on recent advances in measuring territorial control in civil conflicts and using spatio-temporal autoregressive models, I estimate the effect of levels of territorial control and changes to them on the occurrence and intensity of state-based and one-sided violence by governments and rebel groups. Evidence from Nigeria suggests that changes to territorial control are better predictors for violence than current levels of control. This paper contributes to the literatures on civil conflict severity and intensity and spatial analyses of civil conflicts by exploiting variation in territorial control across time and space and showing mechanisms at the grid cell-month-level.