Among a variety of elements, the nearshore wave elimate is considered to be the most important aspect to explain erosion problems along the coast. Oue to the fact that available data is limited, numerical models are valuable and necessary tools to apply. This study is initiated with the main objective to study the nearshore wave elimate of the project area at Nam Dinh coast in Viet Nam by a 2D wave SWAN model in order to investigate the erosion problem along the coast. SWAN (acronym for Simulating of Waves Nearshore) is a third generation wave model, which has been widely applied to simulate wave conditions in many complex field cases in the Netherlands. However, it has never been used in Viet Nam. Thus the achievement of the study may be considered as documentation for future researches of this model in Viet Nam cases. Within this research, at first a SWAN model was set-up based on available data. This includes a sensitivity analysis and model calibration. Then a series of computations were carried out for the application of the model for a mean statistical year. Subsequently, the computed results from the SWAN model were used to identify sediment transport rates. These were compared to actual values in order to investigate whether the 2D wave model can predict the erosion adequately. One of the main results of this study was that the observed sediment transport was qualitatively in accordance with the sediment transport result of the 2D wave model. This emphasized the importance 2D effects to obtain realistic results. In this study, several drawbacks of SWAN have been identified. They include the inconvenient procedure of pre- and post processing in applying SWAN and the convergence problem when activating all physical processes in the model simulations. The study also showed the importance of sufficient wind data in the computational grid. The assumption of uniform wind field in computational grid may lead to inaccurate results in particular with respect to the mean wave direction. For an improvement of the results, the study recommended amongst others that future research should try to generate a variabie wind field in the computational grid.