During the last years, argumentation has been gaining increasing interest in modeling different reasoning tasks of an agent. Many recent works have acknowledged the importance of incorporating preferences or priorities in argumentation. However, relatively little is known about the theoretical and computational implications of preferences in argumentation. paper we introduce and study an abstract preference-based argumentation framework that extends Dung's formalism by imposing a preference relation over the arguments. Under some reasonable assumptions about the preference relation, we show that the new framework enjoys desirable properties, such as coherence. We also present theoretical results that shed some light on the role that preferences play in argumentation. Moreover, we show that although some reasoning problems are intractable in the new framework, it appears that the preference relation has a positive impact on the complexity of reasoning.