This paper is an overview of "Agent-based Computational Economics (ACE)", an emerging approach to the study of decentralized market economies, in methodological perspective. It summarizes similarities and differences with respect to conventional economic models, outlines the unique methodological characteristics of this approach, and discusses its implications for economic methodology as a whole. While ACE rejoins the reflection on the unintended social consequences of purposeful individual action which is constitutive of economics as a discipline, the paper shows that it complements state-of the-art research in experimental and behavioral economics. In particular, the methods and techniques of ACE have reinforced the laboratory finding that fundamental economic results rely less on rational choice theory than is usually assumed, and have provided insight into the importance of market structures and rules in addition to individual choice. In addition, ACE has enlarged the range of inter-individual interactions that are of interest for economists. In this perspective, ACE provides the economist‘s toolbox with valuable supplements to existing economic techniques rather than proposing a radical alternative. Despite some open methodological questions, it has potential for better integration into economics in the future.