The OECD/NRC Boiling Water Reactor (BWR) Turbine Trip Benchmark was analyzed by the code DYN3D and the coupled code system ATHLET/DYN3D. For the exercise 2 benchmark calculations with given thermal-hydraulic boundary conditions of the core, the analyses were performed with the core model DYN3D. Concerning the modeling of the BWR core in the DYN3D code, several simplifications and their influence on the results were investigated. The standard calculations with DYN3D were performed with 764 coolant channels (one channel per fuel assembly), the assembly discontinuity factors (ADF), and the phase slip model of Molochnikov. Comparisons were performed with the results obtained by calculations with 33 thermal-hydraulic channels, without the ADF and with the slip model of Zuber and Findlay. It is shown that the influence on core-averaged values of the steady state and the transient is small. Considering local parameters, the influence of the ADF or the reduced number of coolant channels is not negligible. For the calculations of exercise 3, the DYN3D model validated during the exercise 2 calculations in combination with the ATHLET system model, developed at Gesellschaft für Anlagen- und Reaktorsicherheit for exercise 1, has been used. Calculations were performed for the basic scenario as well as for all specified extreme versions. They were carried out using a modified version of the external coupling of the codes, the "parallel" coupling. This coupling shows a stable performance at the low time step sizes necessary for an appropriate description of the feedback during the transient. The influence of assumed failures of different relevant safety systems on the plant and the core behavior was investigated in the calculations of the extreme scenarios. The calculations of exercises 2 and 3 contribute to the validation of DYN3D and ATHLET/DYN3D for BWR systems.