Since the Three Mile Island accident in 1979, a worldwide effort has been undertaken to understand and to model severe accident phenomena in nuclear reactors in case of a hypothetical loss of core cooling. The international Phebus Fission Product (FP) program, dealing with light water reactor source term research, was initiated in 1988 by the French “Institut de Radioprotection et de Sûreté Nucléaire” (IRSN), and the Joint Research Centre of the Commission of European Communities (CEC), with contribution of Electricité de France, Nuclear Regulatory Commission (USA), CANDU Owners Group and AECL (Canada), Nuclear Power Engineering Corporation and Japan Atomic Energy Research Institute, Korea Atomic Energy Research Institute, Swiss Federal Nuclear Safety Inspectorate “HSK” and Paul-Scherrer-Institute (Switzerland). The aim of the experimental program is to study the degradation phenomena and the behaviour of the fission products (FP) released in the reactor coolant system and the containment building. The program consists of five in-pile tests, performed under different conditions concerning the thermal hydraulics and the environment of fuel rods, in particular the amount of steam (strongly or weakly oxidising atmosphere). Four experiments have been successfully performed so far in 1993 (FPT-0), 1996 (FPT-1), 1999 (FPT-4 in debris bed configuration) and 2000 (FPT-2). The next FPT-3 experiment, including a different neutron absorber material, namely the boron carbide (B4C instead of Ag-In-Cd in the previous tests), which might largely influence both fuel degradation and fission product behaviour, is scheduled by the end of 2004. This paper describes the analysis of the core degradation aspects for the FPT0 and FPT2 tests, using a porous medium geometry to model the thermal behaviour and the relocation of melted materials, with the mechanistic ICARE2 code developed by IRSN. This new approach allowed to simulate the fuel rod swelling for the FPT-2 experiment performed with irradiated fuel (~32 GWd/tU) by varying the initial rod diameter by about 15/20%, which would have an effect on the heat transfers inside the bundle. In this case, the relocation phenomena have been accounted at 2800K corresponding to the interaction between ZrO2, formed during the Zircaloy oxidation period and UO2. This modeling, while exhibiting very encouraging results on FPT-2 test, has been validated against the FPT-0 test where oxidation kinetics and early fuel liquefaction (due to liquid Zry/UO2 interaction) are more important. Though the detailed modeling of such interactions has still to be improved, the ICARE2 3mod1.3 code simulates fairly well the observed fuel degradation, for both the experiments. In the particular case of FPT-0, fuel liquefaction and transition from rod like geometry to molten pool occurred at lower temperature (2450-2650K) largely below the actual melting point of the pure UO2 (3110K). The severe damage observed in the Phebus bundle seems to be due to significant material interactions, initiated by structural materials possibly enhanced by the fuel swelling and fragmentation, and its changes in stoichiometry.