Predictive monitoring is the subﬁeld of process mining that aims to obtain estimations of cases or events features for running process instances. Such predictions are of signiﬁcant value to the process stakeholders. However, state-of-the-art methods for predictive monitoring require the training of complex machine learning models, which is often inefﬁcient. In this paper, we propose an instance selection method that allows sampling training process instances for prediction models. We show that our sampling method allows for a signiﬁcant increase of training speed for predictive monitoring methods while maintaining reliable levels of prediction accuracy.