This paper attempts to define evaluation use and manipulable variables related to it. Approaches to defining evaluation use have included typologies inventorying kinds of evaluation use, threshold criteria defining accepted levels of evidence, construct or operational definitions, and dependence on the perspective of the decision maker. None of these approaches is found to be satisfactory. Within models of evaluation use, the concept is conceived as a dependent variable, and much of the literature constitutes an attempt to identify the factors that act as independent variables in their relation to use. Descriptive models are often structured around components of an evaluation and its setting. Explanatory models, of which there do not seem to be many in the literature of evaluation use, include those developed by A. J. Meltsner (1976), R. D. Brown et al (1984), and G. E. Hall (1981). An effective model of manipulation should have the following characteristics: (1) open-mindedness in defining use as well as non-use and misuse; (2) conceptualization of evaluation use as a process; (3) inclusion of relevant components and their interrelationships; (4) treatment of manipulable and non-manipulable variables; (5) central focus on the interaction of factors; and (6) reduction of variables as much as possible. (TJH)