Insights from leader±member exchange theory and social justice theory were combined to derive predictions about the eects of relative competence and power use by a more powerful other on evaluative and behavioural responses of subordinates. These predictions were tested in two experiments, using a simulated organizational structure. The main results of the ®rst experiment (N 73) indicate that frequency of power use is an important determinant of subordinates' evaluations of the status quo, supporting predictions from leader±member exchange theory. However, in line with a procedural justice explanation, perceived legitimacy of the superior's behaviour and participants' cooperative intentions were determined jointly by power use and relative competence. A second experiment (N 152) more closely investigated behavioural responses. The ®ndings obtained in Experiment 1 were replicated, and as predicted, interaction eects of the superior's power use and competence were found for subordinates' collaborative behaviour and employee turnover. Both these eects were mediated by participants' sense of commitment to their superior. Theoretical and practical implications of the ®ndings are discussed. Copyright # 1999 John Wiley & Sons, Ltd.On a day-to-day basis, managers in organizations are confronted with the question how they may enact their leadership role without frustrating or even antagonizing their subordinates. Inevitably, people with a superior power position in the organization are required to evaluate and if necessary correct the performance of those who work under their supervision. Sometimes supervisors have to take such action without being able to consult their subordinates, or even without having the CCC 0046±2772/99/070843±28$17.50Received 3 Greenberg, 1990;Skarlicki & Folger, 1997). At the same time, people's willingness to behave in a cooperative or altruistic way (their so-called`contextual' performance) is often just as important as their personal task-related activities for organizational success (MacKenzie, Podsako & Fetter, 1991;Podsako, Ahearne & MacKenzie, 1997). Therefore it is crucial to know the circumstances under which such forms of power exertion are likely to harm cooperative eorts from one's subordinates, or may be considered legitimate, so that good working relations between superior and subordinate can be maintained. In this paper we report two studies that examine how subordinates in a simulated organization respond when their eorts are rejected by their superior, in order to predict the consequences this power use may have for future collaboration (see Emerson, 1962). To be able to investigate speci®c theoretical predictions on the eects of power use, we have opted for experimental work in an area where traditionally correlational ®eld studies are done (cf. Tyler & Lind, 1992). This enables us to focus on the causal relations between relevant variables and systematically study their combined and interactive eects, which may help us complement and extend existing insights that are de...