Against the background of enhanced hydraulic loads due to climate change there will be a need for improvement of the flood defence system in the Netherlands in the future. These days there is a growing interest in grass as a dike cover because it is a cheap and a sustainable dike protection. Yet at the moment there is a hiatus in the knowledge on the erosion resistance of grass covers on especially the outer slope. For this reason large scale tests have been performed in the Große WellenKanal (Large Wave Flume) in Hannover in 2008 for both EroGRASS and FLOODsite. With the help of the EroGRASS data the MSc-study presented here aimed to develop a model that describes the initiation of erosion of a grass cover layer on the outer slope by wave attack. The erosion process was investigated first to gain some insight in the failure mechanisms on the outer slope. For wave-induced erosion of grass cover layers on the outer slope two failure mechanisms can be distinguished which can occur independently of each other. Aggregate erosion occurs when the soil is cracked and saturated with water. Uplift pressures can then develop underneath the aggregates shortly after a wave impact and on the surface small aggregates may be lifted and washed away. This eventually results in an erosion hole. Block erosion may occur when impact pressures can penetrate into the soil due to the presence of a large crack or irregularity. The balloon mechanism may then be triggered; at the location of minimum fracture strength a horizontal crack is formed. This crack gradually extends until it reaches a critical size. From this point a large block can instantly erode from the grass cover. For these erosion mechanisms the Wave Impact Pressure Erosion model has been developed, which describes the initiation of erosion of grass covers on the outer slope by wave impact pressures. The basic equation of the WIPE model can be adapted to obtain limit states for aggregate erosion block erosion. The WIPE model was calibrated and verified with the data of the EroGRASS experiments. For aggregate erosion the model behavior resembled the observed progression of aggregate erosion during the experiments after calibration. The WIPE model is considered suitable for the prediction of aggregate erosion of a good quality grass cover. Yet because the grass cover strength is dominated by the grass reinforcement, which decays with depth, the model will require adaptations to make it suitable for grass covers of lower quality. For block erosion the model was calibrated using a parameter that determines the moment of block erosion and a crack growth parameter, which determines the size of the eroded block. As the model was calibrated on merely two characteristic block erosion events, universal calibration factors for block erosion could unfortunately not be found. To obtain more reliable and uniform results for block erosion more data is required.