This chapter focuses on the importance of achieving good generalization capability in predictive process monitoring, ensuring that predictive models can effectively handle new and unseen process instances. It describes the standard evaluation procedure for predictive process monitoring approaches, including the division of event logs into training, testing, and validation sets, as well as the tuning of hyperparameters to optimize model performance. Finally, the chapter explains how trained models are evaluated during the runtime phase using specific metrics to assess their predictive accuracy and robustness.