In the past many viaducts in the Netherlands have been built with precast box beam girders. Precast box beam girders have many advantages compared to other types of girders. One of these advantages is the property of being torsion stiff. Because of boundary conditions some viaducts are skew and not straight. A few examples of these boundary conditions are the limitation due to surroundings or the layout of the (road) intersections. The load path in these types of structures is different compared to a straight viaduct. Increase of torsional moments is expected. Currently there is not a lot of knowledge available about the distribution of forces, mainly concerning shear and torsion, in these types of viaducts. More research is needed to understand the real behavior of these structures and how these can be modelled in a finite element program. The current method of analyzing this type of structures is to use an orthotropic plate model. The shear force and torsional moments cause an increase of the shear stresses in the webs of the girder. This phenomenon cannot be analyzed using an orthotropic plate model because of the difference in cross-sections. In the Eurocode and literature a method approach is provided to be able to translate these forces in a plate model in to shear stresses in the box girder. In this thesis two models are analyzed in DIANA: an orthotropic plate model and a more complex 2,5D shell model. The results of both analyses are compared with each other and the differences are investigated. The analysis is based on a typical box beam viaduct and only the linear elastic stage is considered. The focus is on the comparison of models for the load distributions rather than determining the required reinforcement and capacity calculations. The self-weight and prestressing do not cause torsion in a girder because these loads are applied for the statically determined beam in the factory. The internal forces due to these loads are calculated separately and therefore not inserted in the models. For the Eurocode loading it was found that the maximum value for the longitudinal moment was overestimated by 3,5 percent with the orthotropic plate model when the moment due to the self-weight and the prestressing was taken in to account. The maximum value for the shear stresses was overestimated by the orthotropic plate model 18% when the shear stresses due to the self-weight and prestressing are taken in to account. This led to 23% more shear reinforcement. It is expected that the 2,5D shell model will take two weeks longer to analyze. This means that this will cost 7200 euro for engineering. It is therefore not worth to use the 2,5D shell model to save up money for shear reinforcement (2500 euro) only. It is advised to use the orthotropic plate model for the analysis of new skew box beam viaducts with a skew angle of 60 degrees and the same dimensions as the case study viaduct. This model approach is cost efficient if the design is based on maximum values for the shear stress and the same stirrups are applied over the length of the girder. In case of a reassessment of an existing viaduct where the combination of shear force and torsional moment are governing, the difference of 18% for the shear stress could be decisive in whether the viaduct fulfills the current codes and requirements or not (with the present shear reinforcement). The use of the 2,5D shell model should then be reconsidered.