This research investigates the Wilnis dike failure that occurred in the dry summer of 2003. It is believed that drought played an important role in decreasing the stability of the dike, finally leading to failure. In the reported MSc project the situation during drought is simulated with the help of finite element method programs Plaxis and PlaxFlow in order to get more insight into the conditions that caused the dike to fail. First a forensic analysis based on pictures taken directly after the breach is presented, followed by information about relevant failure cases. The interpretation of the soil investigation performed by GeoDelft after the failure is reviewed subsequently. The field data (Borings and Cone Penetration Tests) are used to define the stratification of the soil. The laboratory data (Triaxial, Simple Shear and Ko-CRS tests) are utilised to determine the soil parameters. With this information a finite element model is constructed in Plaxis. The material model used to simulate the different soil types is the Hardening Soil model. In Plaxis and PlaxFlow 3 different calculation phases are defined. The first 2 have as a purpose to simulate the soil history until the start of drought since this is an important aspect in the further behaviour of soil during drought itself. The third one is the simulated drought period. All the groundwater flow related aspects (groundwater head, excess pore pressures etc.) are calculated by PlaxFlow while the related deformation and stability analyses are done by Plaxis. A parametric study of the groundwater flow by PlaxFlow is performed leading to the conclusion that the upper layer of the soil consisting of Holland peat has to be separated into 2 parts along the middle of the slope of the dike in order to take the different degrees of saturation of the unsaturated soil above the groundwater table, that affect the unsaturated soil weight, into account. Finally 9 different models are composed with varying strengths and permeabilities in order to take into consideration the effect of these parameters. The groundwater flow results are realistic. However the deformation results show irregular unexplainable patterns which are inconsistent with some of the observed data of the actual failure. Furthermore the safety factors are calculated for the end of phase 2 (before drought) and phase 3 (after drought). These results are also very inconsistent and should be discarded.