The current doctoral research is focused on the development and validation of a coupled computational tool, to combine the advantages of computational fluid dynamics (CFD) in analyzing complex flow fields and of state-of-the-art system codes employed for nuclear power plant (NPP) simulations. Such a tool can considerably enhance the analysis of NPP transient behavior, e.g. in the case of pressurized water reactor (PWR) accident scenarios such as Main Steam Line Break (MSLB) and boron dilution, in which strong coolant flow asymmetries and multi-dimensional mixing effects strongly influence the reactivity of the reactor core, as described in Chap. 1. To start with, a literature review on code coupling is presented in Chap. 2, together with the corresponding ongoing projects in the international community. Special reference is made to the framework in which this research has been carried out, i.e. the Paul Scherrer Institute's (PSI) project STARS (Steady-state and Transient Analysis Research for the Swiss reactors). In particular, the codes chosen for the coupling, i.e. the CFD code ANSYS CFX V11.0 and the system code US-NRC TRACE V5.0, are part of the STARS codes system. Their main features are also described in Chap. 2. The development of the coupled tool, named CFX/TRACE from the names of the two constitutive codes, has proven to be a complex and broad-based task, and therefore constraints had to be put on the target requirements, while keeping in mind a certain modularity to allow future extensions to be made with minimal efforts. After careful consideration, the coupling was defined to be on-line, parallel and with non-overlapping domains connected by an interface, which was developed through the Parallel Virtual Machines (PVM) software, as described in Chap. 3. Moreover, two numerical coupling schemes were implemented and tested: a sequential explicit scheme and a sequential semi-implicit scheme. Finally, it was decided that the coupling would be single-phase and isothermal, leaving to future work the extension to more complex cases. The development work itself is presented in Chap. 3, together with a generic consideration of code-coupling issues and the discussion of a few verification cases. After the basic development and verification of the coupled tool, an experiment was devised for its initial validation. The employed experimental set-up, presented in Chap. 4, features a double T-junction, connected to a recirculation loop and instrumented with wire-mesh sensors to measure the concentration of a tracer injected into the flow. The main aim of this experiment has been to challenge the coupled tool with the transport of a tracer in a steady-state flow field. The experimental results, the CFX and TRACE stand-alone simulations, and the CFX/TRACE coupled simulations are compared with each other for validation purposes, as well as for a clear demonstration of the improvements that one can achieve by using a coupled tool. The simulations, at the same time, indicated the occurrence of strong "numerical diffusion" effects in the TRACE simulations, these being found to result from weaknesses in the numerical discretization adopted in the code for the solute tracking equation. Accordingly, as described in Chap. 5, a third-order upwind scheme for the numerical discretization, namely QUICKEST-ULTIMATE, has been implemented in TRACE to replace the original first-order upwind scheme. The mathematical derivation of the new scheme is presented, together with certain verification and validation tests. In particular, the improvements over the original TRACE scheme are shown in the context of the coupled CFX/TRACE simulations of the double T-junction experiment. Finally, a second phase of experimental validation was devised for the coupling. To this end, certain qualification tests for the new FLORIS facility at PSI have been used, as presented in Chap. 6. This second facility features a scaled-down, simplified, two-dimensional vertical slice of a BWR vessel. The aim of this second mixing experiment has been, on the one hand, to challenge the momentum equation coupling in the context of the transport of a tracer in a transient flow field, and, on the other hand, to test the performance of the coupled tool for the case of a more complex geometry. Once again, comparisons have been made between experimental results, CFX and TRACE stand-alone simulations, and CFX/TRACE coupled simulations, employing the QUICKEST-ULTIMATE discretization where possible. As before, it is clearly demonstrated that the coupled tool yields much better results than the stand-alone codes. Furthermore, it has been found to be sufficiently robust for being extended to more advanced applications, such as the analysis of PWR transients in which strong reactivity feedback effects occur in the context of complex coolant flow phenomena.