This thesis focuses on performance, reliability and degradation in solid oxide fuel cells (SOFC), which currently represent the three key technical challenges for the deployment of this technology. By its development of dedicated modeling and experimental tools, this thesis offers a new access to, and comprehension of, local electrochemical performance, degradation and reliability. The experimental tools developed have, for the first time, allowed in-situ measurements of local performance degradation with time, and this in a real SOFC prototype. Giving access to local electrochemistry on 18 measurement points in a repeat-element of 200 cm2 active area, the spatial distribution of the electrochemical reaction, as well as its degradation with time, could be monitored and analyzed. The intrinsically local character of degradation could clearly be revealed, a point of central importance for future investigations on stack degradation and for the interpretation of post-experiment analyses. Using impedance spectroscopy, it was possible to identify the affected electrochemical processes and to study the spatial distribution of their degradation. The result was put in relation with post-experiment analyses, allowing to identify pollutants on the air side as major source of degradation. To understand the highly coupled phenomena leading to performance, degradation and reliability issues, a 3D computational fluid dynamic model (CFD) was developed. Based on the key idea to include the non-ideal properties of the used components and materials in the model, it was possible to obtain an excellent match between experimental observations and modeling outputs. Besides the identification of performance limitations, one result of crucial importance was in addition obtained by the identification of the principal cause of failure for the prototypes tested in the laboratory. The model revealed the presence of detrimental local redox-cycling of the cells upon changes of the operating point, as a result from an inadequate combination of slightly porous seal materials and certain aspects of the stack construction. This analysis, validated by experimental observations, led to solutions permitting an important gain in efficiency and reliability. Based on the identified and analyzed performance limitation, degradation and failure sources obtained from the tools developed in this thesis, two stack prototypes were successively designed, manufactured and tested in collaboration with the industrial partner HTceramix-SOFCpower. Starting from a predecessor design limited at 250 Wel and an electrical efficiency (LHV) inferior to 40%, the first of the designed prototypes attained a power output of 1.1 kWel (72 cells of 50 cm2), as well as a maximal efficiency of 53% in short stack configuration. The second designed stack, which represents the major achievement of this thesis, had, at the time of writing, reached a power output of 1.84 kWel and a maximal efficiency of 53% in a 20-cell stack configuration (200 cm2 cells). Both results were obtained using dilute hydrogen as fuel; in other words, future operation on reformed natural gas should lead to an electrical efficiency exceeding 60% (LHV). With the successful resolution of the main failure source and a demonstrated gain in efficiency, the chosen design iterations confirmed the predicting capabilities and the accuracy of the CFD model for a design towards the mandatory reliability and the high performance expected from SOFC stacks. Finally, the degradation issue, which was found to be strongly correlated with pollution from different sources, was addressed in a prospective study showing the capability of the CFD model to predict the internal generation, transport and deposition of pollutants inside of a stack. The good match obtained with experimental observations supports therefore the development of such types of models, both for model-based diagnostics and for future design iterations.