Nuclear power is nowadays in the front rank as regards helping to meet the growing worldwide energy demand while avoiding an excessive increase in greenhouse gas emissions. However, the operating nuclear power plants are mainly thermal-neutron reactors and, as such, can not be maintained on the basis of the currently identified uranium resources beyond one century at the present consumption rate. Sustainability of nuclear power thus involves closure of the fuel cycle through breeding. With a uranium-based fuel, breeding can only be achieved using a fast-neutron reactor. Sodium-cooled fast reactor (SFR) technology benefits from 400 reactor-years of accumulated experience and is thus a prime candidate for the implementation of so-called Generation-IV nuclear energy systems. In this context, the safety demonstration of SFRs remains a major R&D related issue. The current doctoral research aims at the development of a computational tool for the in-depth understanding of SFR core behavior during accidental transients, particularly those including boiling of the coolant. An accurate modeling of the core physics during such transients requires the coupling between 3D neutron kinetics and thermal-hydraulics in the core, to account for the strong interactions between the two-phase coolant flow and power variations caused by the sodium void effect. Models for the representation of sodium two-phase flow are not present in most of the thermal-hydraulics codes used currently, and these have specifically been focused upon. The particular contributions of the present research are: (1) implementation of sodium two-phase flow models into the thermal-hydraulics code TRACE, which forms part of the FAST code system at PSI and, as such, can easily be coupled to the spatial neutron kinetics code PARCS; (2) validation of the TRACE sodium single- and two-phase flow modeling using out-of-pile sodium boiling experiments; (3) validation of the coupled TRACE/PARCS code system on the basis of experimental reactor data; and (4) application of the developed new tool for the core behavior analysis of an advanced SFR during a transient with boiling onset. The extension of the TRACE code, previously limited to the simulation of single-phase sodium flow, has been carried out through the implementation of equations-of-state and closure relations specific to sodium. A review has first been performed of the models available in the open literature for the representation of the interfacial and wall-to-fluid transfer mechanisms. The different correlations have then been implemented as options in the extended TRACE code. From the validation study carried out, it has been possible to recommend a set of models which provide satisfactory results, while considering annular flow as the dominant regime up to dryout and a smooth breakdown of the liquid film after dryout onset. The validation of the extended TRACE code has been achieved through the successful simulation of out-of-pile experiments. A review of available sodium boiling test data has first been carried out, and complementary tests have then been selected to assess the quality of the different physical models. These tests, performed in the 1980s, include the study of the pressure drop and cooling limits under quasi steady-state conditions, as well as the simulation of a loss-of-flow transient. Sensitivity analyses, using the specifically implemented correlations, have enabled one to identify the most pertinent physical parameters and to define, for each, the most appropriate model. Usage of the set of models thus selected has demonstrated the capacity of the extended TRACE code to predict, with satisfactory accuracy, the main thermal-hydraulics characteristics such as the single- and two-phase pressure drop and heat transfer, as also the characteristic quantities describing the sodium two-phase flow, e.g. boiling inception, void fraction evolution and expansion of the boiling region, pressure evolution, as well as coolant and clad temperatures. The natural convection test conducted in 2009 in the Phenix reactor has been used to validate TRACE single-phase sodium flow modeling. This represents the first international benchmark exercise conducted on the basis of actual SFR experimental data. Data from the Phenix test have additionally been used as basis for the validation of the FAST code system as a whole. Analyses based on a point-kinetics TRACE model and on coupled TRACE/PARCS 3D-kinetics modeling have enabled an in-depth understanding of the transient behavior of a sodium-cooled fast reactor core, as well as the identification of potential improvements in the FAST code system. The experimental power evolution could be satisfactorily reproduced within the measurement uncertainties with both models, and the detailed analysis of the core neutronics has enabled one to define the most important reactivity feedbacks taking place during the considered transient. In the final part of the thesis, the developed tool has been applied to the simulation of a hypothetical, unprotected loss-of-flow event for one of the European SFR (ESFR) core concepts. This study has demonstrated the new calculational tool's capability to adequately simulate the core response through the modeling of single- and two-phase sodium flow, coupled to 3D neutron kinetics. Thereby, the space-dependent reactivity feedbacks, such as the void and Doppler effects, have been shown to be modeled accurately. The resulting analysis has shown the ability of the TRACE/PARCS modeling to predict the expansion of the boiling region and calculate the resulting feedbacks, as well as to predict the interactions between parallel boiling channels. This first-of-a-kind study has provided detailed results for the thermal-hydraulics and neutronics parameters during the pre-severe phase of the simulated accident, thus allowing a comprehensive understanding of the core behavior during such transients. In brief, the present research has led to the development of a key calculational tool for SFR safety analysis. A first application of the new tool has demonstrated its potential for usage in SFR design optimization aimed at enhanced safety.