This paper examines the hypothesis that a change in the elite increases the probability of armed conflict onset. The paper combines the newly constructed WhoGov dataset on cabinet members, with UCDP data on armed conflicts and rebel group. It analyzes changes in the retention rate among cabinet members and average years in cabinet, and how they affect armed conflict – measured as armed conflict onset and rebel group onset. A linear regression model, with country- and year fixed effects, in addition to time-varying control variables, is applied to test the main hypothesis. In addition, three heterogeneity tests are conducted. First, the paper examines whether there are any differences between democracies and autocracies. Second, it examines the differences between changes in cabinet size (widening or shrinking). Third, it analyzes if the effects are bigger for large changes in the elite. Four main conclusions are reached in the paper. First, the results indicates that a decrease in retention rate increases the probability for both conflict and rebel onset. The magnitudes of the estimated effects seem to be sizeable. They are larger in conflict onsets than rebel onsets, in relation to their respective means. Second, for armed conflict onset, the results are driven by autocracies. Third, the probability for armed conflict is larger following a large change in the elites (retention rate). Fourth, no difference in changes to cabinet size is observed.