Many philosophers of science believe that a theory's successful prediction of the phenomena, also known as novel success, is a more impressive kind of empirical success than a theory's accommodation of already known phenomena. This chapter argues that none of the proposed rationales for predictivism, as this view has also been called, is convincing. More specifically, standard notions of temporal novelty, use-novelty, novelty as parameter fixing, comparative novelty, and others are all rejected as problematic. Given that realists usually base their commitment on whether or not a theory manages to produce novel success, it is argued, standard realist responses to the Pessimistic Meta Induction, such as a reduction of Laudan’s list and the realist’s divide et impera move, are at risk of being undermined.