Under what conditions will a state repress its citizens? The literature examining human rights violations lacks consensus over exactly how repression and dissent are interrelated. I argue that contradictions have arisen because scholars have not derived expectations consistent with modeling three common assumptions: (1) dissent and repression are causally interrelated (2) states and groups are in conflict over some policy or good and (3) authorities repress to remain in office. I develop a formal model based on these principles, and I predict that changes in the same independent variable can have divergent effects on the onset and severity of repression. Using coded event data for all states from 1990 to 2004 and a two-tiered estimator, I find that increases in executive job security decrease the likelihood that repression will occur in the first place, but increase the severity of observed violations.