Many experiments show that threats to attitudinal freedom create reactance, but the underlying dynamics of reactance-based disagreement have not received much attention. The present experiments identified two paths from threats to disagreement. In one path, threats to attitudinal freedom directly motivate disagreement; in the other, negative cognitive responses mediate the threat's effect on disagreement. Two experiments demonstrated the causes and consequences of each path from threat to persuasion. When a communicator threatened freedom at the beginning of the message, unfavorable cognitive responses (counterarguing, negative perceptions of the source's credibility) fully mediated the effect of threat on disagreement. When the threat appeared at the end of the message however, threat had a direct, unmediated effect on disagreement (Experiment 1). The two paths had different consequences for sleeper effects: disagreement rooted in negative cognitive responses persisted, whereas disagreement directly motivated by the threat declined when the threat was removed (Experiment 2). Implications for reactance and for threat-based sleeper effects are discussed. Copyright # 2006 John Wiley & Sons, Ltd. Reactance theory (Brehm, 1966;Brehm & Brehm, 1981;Wicklund, 1974), one of the most widely-studied theories of resistance to social influence, proposes that the motivation to maintain personal freedoms creates resistance to persuasion. According to the theory, people are motivated to restore a freedom when they perceive that it has been threatened or eliminated. Several freedoms are involved in the domain of attitude change (Brehm & Brehm, 1981, chap. 6). People feel free to hold particular attitudes, to change their attitudes, or to avoid committing to any position (e.g., Wicklund & Brehm, 1968;Worchel & Brehm, 1970). When these freedoms are threatened, people experience reactance. Attitude change is one way to restore threatened attitudinal freedom. If a communicator threatens one's freedom to disagree, then the freedom to disagree can be reasserted by disagreeing (Wright, 1986). This pattern is the well-known 'boomerang effect'-the recipient moves away from the position advocated by the communicator. Chaiken (1993) pointed out that the inner dynamics of reactance were not well understood:The issue of how reactance, the negative emotional state that ensues when freedom is threatened or eliminated, influences the processing of information remains largely unexplored. Researchers who have used reactance theory to generate predictions or to explain obtained persuasion findings have rarely included measures that may provide evidence of subjects' cognitive processing . . . .This omission is not surprising given that most research on reactance predated attitude researchers' contemporary preoccupation with underlying cognitive processes (p. 571). Brehm and Brehm (1981), in discussing unresolved issues in reactance research, makes a similar point. After noting the evidence for threat-induced boomerang effects, they ask,...