A large part of the Vietnamese population is living close to the coast. The sea is an essential source for supporting lives. Coastal flooding, on the other hand, threatens the people, land and infrastructures every year. Some hundreds years ago, the first sea dikes were built in Thai Binh and Nam Dinh provinces. In the north, the coastline is now protected with more than 700 km of estuary and sea dikes. Being directly exposed to waves and currents, these dikes must be sufficiently stable and durable to function as coastal defences. The present thesis contributes to better understand how a dike performs under overtopping during a storm surge. The research limits itself to a slope that is covered with grass on the landward side. Therefore, this slope is mainly attacked by the overtopping flows. The research mainly explores four issues concerning a) damage pattern of grass dike slopes in Vietnam, b) how to quantify overtopping flows, c) how roots strengthen soil within a grass cover and d) classify and model damage to slopes due to overtopping. First, a wave overtopping simulator was used to test the strength of some grass covered slopes in Vietnam. The simulator was designed to generate overtopping flows at real dikes in normal weather conditions. The tested slopes could withstand a mean discharge varying from 20 to 100 l/s per metre of dike length during several hours. Damage usually starts at bare spots, at the inner toe, at the transition between different materials, and around objects (e.g., big trees). These features reduce the strength of a dike slope and therefore should be avoided as much as possible. Second, the thesis established formulas to quantitatively estimate overtopping flows. To this end, the research re-analysed an available set of data. The flow velocity and the water-layer thickness are related to the corresponding run-up level and overtopping volume. Within their ranges of application, the formulas can predict the flows in front of the simulator and on the dike slope relatively well. Third, the thesis investigated how roots strengthen soil on Vietnamese dike slopes. Measurements were carried out with four grasses Bermuda, Carpet, Ray and Vetiver to determine the diameter, the volume, the weight and the number of roots; and how strong they are. About 60 to 80% of roots have a diameter varying between 0.1 to 0.2mm. The volume, the weight are proportional to the number regardless of the size. About 60 to 90% of the root amount are present within the top 10 cm under the soil surface. There are less than 10% under 20 cm. A root thread can resist a pulling force of several Newton and this value often increases when grass becomes older. Formulas are developed to estimate the reinforcement to soil where roots are embedded in. The soil strength depends directly on the number (density) and the strength of roots. Fourth, the findings achieved earlier are integrated to investigate how a grass covered slope behaves when being attacked by overtopping flows. Based on the observations with the simulator tests, damage is classified into different manners depending on grass, material components and corresponding layer thickness. Erosion usually starts at weak spots so the thesis examined their spatial distribution along a dike stretch. The resistance of a grass cover can be estimated with two approaches. The first one assumes a critical velocity depending on the soil cohesion and the apparent cohesion induced by roots. The cover is damaged when the flow velocity exceeds the critical value. The second one solely considers a total tensile strength produced by the intact roots. The flows exert force to gradually break root by root, and the Root Bundle Model was developed to assess this process. Calculations show that older grass with a greater density and stronger threads of roots increases the resistance. By addressing these four issues above, the thesis has achieved its main research objective. By and large, it provides more profound insights into the damage of a grass covered slope due to overtopping. Besides, the study proves the potential strength of grass in protecting a dike slope. This strength would increase proportionally to the development of grass. Notably, a mean discharge appears to indicate partly the destructive effect of overtopping. Therefore, one should be aware of the simplification when using this value to design and appraise a sea dike. Last but not least, high dikes with no overtopping are expensive to obtain in developing countries like Vietnam due to the budgetary constraints. As a matter of fact, sea dikes might be lower and therefore cheaper when certain rates of water are allowed by taking into account the resistance of the applied grass.