Participants were asked to put themselves in the position of one of three persons who differed in the amount of power they had in a small work unit. Subsequently, they could allocate points on a power scale to themselves and the two others, and thus, change the power positions and the power distances between the positions. The least powerful individuals had the strongest tendency to increase their power. They wanted to reduce the power distance to the person in the higher position more than the power distance to the person in the middle position. The most powerful wanted to increase the power distance to the person in the middle position but not the power distance to the least powerful. Most results were consistent with social comparison theory and contrary to power distance theory. Because the dependent measures were derived from the social value orientations model, the scope of this model has been expanded. Copyright # 2002 John Wiley & Sons, Ltd.Many relationships, groups, and organizations are stratified because people differ in how much power they have. Bosses usually have more power than foremen and foremen usually have more power than workers. Often people want to change the power differences. Understanding power dynamics is an important element of understanding the functioning of relationships, groups, and organizations. In this article we will focus on two theories which explain why and how people want to change power differences. One theory, power distance theory, has been developed especially for understanding people's preferences to change power differences. The other, social comparison theory, is a more general theory that can be applied to preferences for changing power differences. Because some of the hypotheses which can be derived from the two theories are contradictory, an experiment will be described in which some hypotheses were refuted and others supported. In power dynamics, several tendencies-e.g. a tendency to change one's absolute power and a tendency to change one's relative power (compared to that of another person)-are entangled. To disentangle these tendencies methods were borrowed from research on social value orientations. The central aim of this article is to evaluate the two theories with respect to power dynamics. Further, we hope to demonstrate that the social value orientations paradigm is a useful framework to study interpersonal phenomena such as social power.