In this chapter, the authors present methods that are useful for a detailed examination of both overall and instance-specific model performance. In particular, they focus on graphical methods that use residuals. Thus, in the chapter, the authors are not aiming at being exhaustive. Rather, their goal is to present selected concepts that underlie the use of residuals for predictive models. To evaluate the quality, the people should investigate the "behaviour" of residuals for a group of observations. For independent explanatory variables, it should lead to a constant variance of residuals. Diagnostic methods based on residuals are a very useful tool in model exploration. They allow identifying different types of issues with model fit or prediction, such as problems with distributional assumptions or with the assumed structure of the model.