Soil erosion models give scientific support for soil conservation and water management over a wide range of technical and research disciplines at different spatial and temporal scales and in many different environments. From the first empirical procedures that appeared as early as the 1930’s to current models, all predictions have needed validation against quantified rates of soil loss, deposition and sediment yield. Nowadays, soil erosion models are ambitious, mathematically complex and often GIS based, and many different processes of soil loss and sediment transport represented in their equations. Predicting models are widely used in a huge variety of environments and scales to predict the hydrological and soil erosion response to different management and climatic scenarios and in some cases the models can be downloaded as executable files. The different temporal and spatial scales of application of the models make validation procedures specific to each model and its characteristic scales. Validation data can be acquired, for example, from gauging stations at the outlet of a catchment, from reservoir bathymetry, from lake sediment cores, from radionuclides (net rates of soil redistribution, sediment chronology, finger printing), from direct measurements with erosion pins and from monitored experimental plots. In this session we seek to discuss the current state of validation procedures, the associated levels of uncertainty and future challenges. The accuracy, shortcomings and advantages of each technique as well as the different temporal and spatial scales of application of each approach will be discussed in this session.