Runout analysis is a key component of landslide risk assessment and management, and a range of empirical and numerical methods for analyzing runout are available. Significant advances have been made in the development of numerical runout models over the past decade, particularly with respect to three-dimensional modeling capabilities. As demonstrated in a recent model benchmarking workshop, most modern numerical models are able to simulate the bulk characteristics of typical real landslides. On the other hand, progress has been slower in developing suitable methodologies for selecting input parameter values for prediction. Ideally, these methods should fit within the probabilistic framework of quantitative landslide risk assessment, allowing users to estimate the spatial probability of impact and associated hazard intensity throughout the runout zone. Recent work on advanced model calibration techniques has attempted to address this need. Simple probabilistic empirical–statistical techniques can be extended to numerical modeling applications and provide a useful reference point for these discussions.