This thesis focuses on the degradation pathways occurring in composite solid oxide fuel cell (SOFC) cathodes based on lanthanum-strontium manganite (LSM), combining modeling at the electrode level with experimental data. LSM composite electrodes are one of the most promising candidates for SOFCs cathodes, yet life-time is currently one of the limiting factors for the deployment of this technology. The combination of modelling tools and analysis of experimental data permitted to achieve a better comprehension of the mechanisms controlling the performance loss during operation and a quantification of the degradation. By acquisition and formalization of knowledge of degradation processes, this thesis offers new tools to predict the life-time of a cell, hypothesize technological solutions for limiting the performance losses, and evaluate the impact of such solutions. A new experimental strategy for SOFC button cell testing has been developed and validated during this work, namely the simultaneous operation of several cathode segments on the same cell support. This allowed, on the one hand, the production of reproducible and reliable data for the evaluation and investigation of degradation processes and, on the other hand, the possibility of rapid identification of experimental problems affecting one segment. The approach has been validated both experimentally and theoretically – through finite element calculations. The vast majority of the experimental results contained in this work has been obtained using this testing configuration. Furthermore, investigation techniques with unprecedented resolution in the SOFC research field have been developed in collaboration between EPFL and external partners, namely a fast Cr quantification in operated cells, and 3D non-destructive reconstruction of cathode microstructures by X-ray computed tomography, giving new instruments for the study of SOFC performance and degradation. Finally, a set of data has been gathered concerning the reactivity of LSM-based systems, aging composite pellets for different times and temperatures and analyzing them with Rietveld-XRD, in order to map and assess the loss of performance caused by the formation of insulating phases in the electrode. A steady-state electrode model present in literature has been selected and improved for simulating the performance of composite cathodes in presence of degradation phenomena, converting it into a time-dependent model. The first phenomenon that has been integrated is the variation of the microstructure of the composite electrode, allowing the prediction of variation in performance for a number of case studies: coarsening of only one phase, coarsening of both phases, coarsening of both phases with variation of porosity. Experimental results obtained by microstructural analysis and electrochemical characterization of cells aged for different times were compared to the simulations of the correspondent case study, showing that the morphological variation of the anode electrode in the early cell operation (first few hundreds hours) is well predicted by the time-dependent model. No cathode morphological variation on this time scale could be detected. Another degradation pathway analyzed was poisoning by Cr species. The phenomenon has been modelled assuming an overpotential-driven deposition of Cr blocking species on the electrode active sites. Validation has been performed comparing the simulations to Cr pro les obtained experimentally from cell tested for relatively short time (1000 h); one conclusion reached was that increased electrode thickness has a beneficial effect in limiting Cr poisoning. The simulation was extended to operation times on the order of several 10 kh, difficult to achieve experimentally, describing the progressive deactivation of the electrode material. The model provides an estimation of the expected life-time of an electrode and predicts quantitatively the performance loss during time; this allows individuating technological solutions to hinder Cr poisoning, and the evaluation of the impact of such solutions. In particular, it has been found that the conductivity and amount of the electrolyte phase are critical, and that a three-fold decrease of the degradation related to Cr poisoning is expected while passing from 50%LSM/50%YSZ (standard composition for most LSM based cathodes) to 42%LSM/58%ScSZ electrodes (new composition). Finally, it has been found that cathode composition may have an indirect effect on the amount of deposited Cr: experimental evidences supporting a Mn-catalyzed deposition of Cr species have been found, suggesting that the presence of excess Mn in the cathode system can be detrimental.