Instrumentality or expectancy theories have recently been used to predict the effort, performance and satisfaction of employees in various organizational settings. The theory states that behavior can be predicted from the probability of the act leading to some outcomes multiplied by the evaluation of these outcomes. Much of the research using these ideas was originally conducted in the area of decision making. The paper presents a critical analysis of the attempts to generalize these ideas from the decision area to other areas of organizational behavior (e.g., effort and satisfaction). Special reference is made to the conceptual and methodological problems involved in the measurement of the theory's two components. (Author)