Theoretical virtues of scientific theories, also known as epistemic values or cognitive values, are taken by many philosophers of science as crucial factors in theory choice. This dissertation investigates two main issues: what role do these theoretical virtues (e.g. internal consistency, external consistency, empirical adequacy, accuracy, simplicity, explanatory power, non-ad hocness, predictive power, unification, fertility, and scope) play in theory choice in science? Can they be used, fruitfully and justifiably, in metaphysics? In Chapter 1, I argue that in the context of scientific theory choice, there is no algorithm, no set of rules, for choosing the theory that best exemplifies theoretical virtues over its rivals. In Chapter 2 I argue that all theoretical virtues are constituents of the goal of scientific inquiry and scientists aim to formulate theories that exemplify the best possible balance of all theoretical virtues. Then, by examining the centrality of theoretical virtues to our understanding of scientific rationality, the (approximate) truth of successful scientific theories, and scientific progress, I argue that theoretical virtues are fundamental for and constitutive of the practice that is known as science. Chapter 3 studies two classic accounts of inferring explanations from evidence—i.e. Charles Sanders Peirce's theory of abduction and Peter Lipton's account of inference to the best explanation—to see what role these philosophers propose for theoretical virtues in explanatory reasoning and how their accounts are connected. I argue that differences between Peircean abduction and Lipton's inference to the best explanation can be well understood in terms of two developments in the history of philosophy of science, i.e. the Duhem-Quine thesis of underdetermination and Reichenbach's distinction between the context of discovery and the context of justification. Chapter 4 aims to see whether theoretical virtues can be used justifiably and fruitfully in metaphysical theory choice. Relying on the results of previous chapters, I argue that under certain conditions, theoretical virtues can play such a role in metaphysics. In particular, theoretical virtues can be used in metaphysical theory choice where metaphysics aims to explain aspects of the world as described by our best scientific theories.