This essay considers the place of mechanisms in ancient theories of science. It might seem therefore to promise a meager discussion, since the importance of mechanisms in contemporary scientific explanation is the product of a revolution in scientific thinking connected with the late Renaissance and its mechanization of nature. Indeed the conception of astronomy as devoted merely to "saving the appearances" without reference to the physics of planetary motion might seem an instance of ancient science vigorously rejecting mechanisms. This fact should suggest to us that there is no simple truth about the role of mechanisms in science that is not relative to a particular strategy or historical moment in the development of a scientific tradition. It should also remind us of a parallel question concerning scientific theory, the question now expressed in the issue of realism and instrumentalism. A discussion of Aristotle's views on scientific explanation in the Posterior Analytics, and particularly his concept of what is prior and better known by nature, shows the ways in which Aristotle resists a crude choice between realism and instrumentalism. Early modern theories of science were committed to realism, and the notion of mechanisms was important to that commitment. Part of the force of mechanisms was that they were thought to reveal the activities by which phenomena are truly brought into being. In this way they often served early philosophers of science by promising a realistic science, able to discover the actual mechanisms by which phenomena were brought about in a universe increasingly viewed as mechanistic. Mechanisms, then, are of critical importance if one is a realist, but of considerably less importance if one is an instrumentalist. Since Aristotle, at least, was neither a realist nor an instrumentalist, his view might be thought to be typically Aristotelian: in some ways they're important, in some not. A glance at some contemporary instances suggests similarly that there can be no general view from a theoretical perspective of the status of mechanisms. Their place is dependent on specific features of specific projects of scientific research and explanation.