Within the framework of the European project CLASH phenomena related to wave overtopping are studied. In this report a method to predict wave overtopping at a wide range of coastal structure types is given. The method is based on Neural Network modelling of the mean wave overtopping discharge. The method can be applied for coastal structures such as dikes, rubble mound breakwaters and caisson structures. The predictions by the Neural Network go together with information on the reliability of these predictions (confidence levels). The quality of the predictions strongly depends on the quality of the database on which the Neural Network is based. The data base used here consists of about 8400 test conditions. These data mainly originate from small-scale tests from many different laboratories. The results show that rather accurate predictions can be obtained with the Neural Network