This paper compares the Best-Estimate Plus Uncertainty (BEPU) methodology with the Conservative Bounding methodology for design-basis-accident analysis. Calculations have been performed with TRACE [for thermal-hydraulic (TH) system calculations] and PARCS [for neutron-kinetics (NK) modeling] under the SNAP platform. DAKOTA is used under the SNAP interface for uncertainty and sensitivity analysis. A simplified three-dimensional (3-D) neutronics model of the Ascó II nuclear power plant is used as the core kinetic model. The TH model is a one-dimensional representation of the primary and secondary systems except for the vessel, which is represented by a 3-D VESSEL component. The design-basis transient selected for the comparison is a main steam line break (MSLB) in a pressurized water reactor. This transient is characterized by space-time effects and requires coupled 3-D kinetics and TH modeling, especially for the recriticality scenario. The comparison methodology is as follows. Once the models are created, a best-estimate base case calculation is performed. The model is modified by selecting the most important parameters and assigning conservative values to them in order to obtain a conservative calculation. Several parameters are modified in this conservative way. These parameters are then perturbed in BEPU calculations. At the end, a comparison is made between results obtained in the conservative calculation and the BEPU methodology, respectively. As a general conclusion the BEPU method has been successfully illustrated in a coupled 3-D kinetics and TH system. Also, the study is an effective test for the adequacy of nodalizations for the neutronic and TH utilized codes. The BEPU methodology gives more margins, which allows for higher operational flexibility of the plant. The results of the BEPU methodology help improve the plant economics while meeting the safety standards. As a conclusion, the BEPU methodology has been successfully tested in NK-TH calculations. Narrow margins between the upper and lower BEPU cases are a consequence of the few perturbed parameters chosen and the transient boundary conditions.