The applicability of the open source model XBeach was tested for pocket beaches in this paper. First, schematized model setups have been used to assess the performance of XBeach qualitatively. It was found that adjustments of the initial bed were required to obtain the expected hydrodynamic conditions. The default sediment advection calibration factors had to be increased to obtain the expected morphological response of the pockets (stationair: 0.15, surfbeat: 0.30). The results have shown that the surfbeat mode outperforms the stationary mode based on a qualitative assessment. It was furthermore found that wave-current interaction produced unreliable results and should be avoided when modelling pocket beaches using XBeach. Non-hydrostatic simulations produced significantly more sedimentation behind the headlands due to the inclusion of diffraction. This mode was found to produce unexpected erosion at the upper shoreface however, indicating the inapplicability of the sediment transport formulas in non-hydrostatic simulations. Expansion of the classic pocket beach into two adjacent pocket beaches showed that these systems have to be treated as a complete system if interactions between these pockets are expected (separated by a salient). It was found that an individual assessment is adequate if a tombolo separates the pockets. Second, the morphological developments between the year 2000 and 2015 at Tanjong Beach (Singapore) were modelled using XBeach. For the modelled bed level changes, which were compared to bathymetric measurements, a ‘Good’ Mean-Squared Error Skill Score of 0.31 was found. In the prepared model, the obtained sediment advection calibration factors from the first part were used and model calibration was not performed. Based on the schematizations and the case study performed in this study, XBeach was found to be applicable for long term process-based modelling of pocket beaches.