This chapter examines the core principles of statistical learning theory. It begins by analyzing the concept of consistency in empirical risk minimization (ERM), a foundational element underlying most supervised learning algorithms. The discussion then turns to worst-case scenarios in learning theory, highlighting the significance of robustness and generalization. Building on the analysis of consistency, the chapter introduces structural risk minimization (SRM), the second key principle in supervised learning, which has driven the development of advanced machine learning models. Essential concepts such as the VC dimension and the growth function are introduced to clarify how model capacity and generalization are assessed. Finally, the chapter details methods for upper-bounding a classifier’s generalization error, outlining the necessary assumptions and theoretical tools for this analysis.