Fitted models must be thoroughly checked to ensure that they provide a good description of the experimental data and, conversely, that the data adhere to the basic assumptions that the models make. Without this check, we cannot be sure that the fitted models can provide a reliable answer to the research questions we intend to ask. In this chapter, a checking routine is proposed for model residuals based on several types of graphs that are easily accessible in R, together with a few simple formal statistical tests for normality and homoscedasticity. Variance stabilizing transformations are introduced as a swift remedy for model violations, and guidance is provided on how to select the most suitable type of transformation, based on the widely used Box-Cox method.