Vigorous and controversial, this book develops a sustained argument for a realist interpretation of science, based on a new analysis of the concept of predictive novelty. Identifying a form of success achieved in science--the successful prediction of novel empirical results--which can be explained only by attributing some measure of truth to the theories that yield it, Jarrett Leplin demonstrates the incapacity of nonrealist accounts to accommodate novel success and constructs a deft realist explanation of novelty. To test the applicability of novel success as a standard of warrant for theories, Leplin examines current directions in theoretical physics, fashioning a powerful critique of currently developing standards of evaluation. Arguing that explanatory uniqueness warrants inference, and exposing flaws in contending philosophical positions that sever explanatory power from epistemic justification, Leplin holds that abductive, or explanatory, inference is as fundamental as enumerative or eliminative inference, and contends that neither induction nor abduction can proceed without the other on pain of generating paradoxes. Leplin's conception of novelty has two basic components: an independence condition, ensuring that a result novel for a theory have no essential role, even indirectly, in the theory's provenance; and a uniqueness condition, ensuring that no competing theory provides a basis for predicting the same result. Showing that alternative approaches to novelty fall short in both respects, Leplin proceeds to a series of test cases, engaging prominent scientific theories from nineteenth-century accounts of light to modern cosmology in an effort to demonstrate the epistemological superiority of his view. Ambitious and tightly argued, A Novel Defense of Scientific Realism advances new positions on major topics in philosophy of science and offers a version of realism as original as it is compelling, making it essential reading for philosophers of science, epistemologists, and scholars in science studies.