Subjects rendered judgments regarding the power of the participants in a series of conflictual circumstances where an adversary threatened a target. These situations manipulated four independent variables: (a) the adversary's capacity to damage the target's interests, (b) the adversary's probability of actually attacking, (c) the target's ability to block the impending attack, and (d) the target's capacity to retaliate. Results showed that all of the independent variables affected the subjects' judgments of the adversary's power, while three of them (damage, blockage, and retaliation) affected judgments of the target's power. Differences in the predictive equations for judgments of adversary power and target power were noted, and a theoretical model was formulated to explain these differences. This model, cast in terms of the patterns of control exercised over valued outcomes, sharpened the focus on remaining issues in power perception.