It has often been proposed that people are intrinsically motivated to gain or increase power over others. We argue that theoretical underpinnings of such a claim are lacking. Moreover, empirical support for this claim is more convincingly explained by strivings to increase one's sense of agency (personal power) by decreasing dependence on others, rather than by strivings to increase power over others (social power). In two experiments, we directly tested the explanatory value of the personal power concept. In Experiment 1, participants performed a decision-making task, together with a (simulated) other person. The power of the two persons over each other was manipulated orthogonally by varying the control they had over each other's decisions. As expected, the participants mostly increased their personal power, by decreasing their dependence on the other person's power. They did not increase their social power but even decreased it when they were very superior themselves. Comparable findings were obtained in Experiment 2, in which participants interacted with another person whose decisions conflicted with those made by the participant. Copyright © 2006 John Wiley & Sons, Ltd.