One of the strategies to mitigate the consequences of a severe accident is to stop the progression of molten materials (corium) within the vessel by external cooling of the vessel. This strategy is called in-vessel retention (IVR). The safety demonstration of IVR requires the accurate evaluation of the thermal load imposed on the internal surface of the vessel by the corium debris. During the progression of the accident, solid debris may be generated initially in the lower plenum because of fragmentation in water and, after melting of those debris, a (possibly stratified) molten pool would be formed. The consequences of the thermal and chemical interactions of the corium with the vessel are difficult to predict because there are still uncertainties on the application of existing results to models for reactor cases. These include scaling effects of course, but also the influence of transient processes occurring simultaneously: oxidation, dissolution, melt progression and/or stratification. In order to be able to predict the heat fluxes through the RPV wall, it is necessary to establish an overall view of the possible transient evolutions of corium and determine which are the most probable ’steady-state’ corium configurations after debris melting. The CORDEB program studies corium phenomena taking place at different stages of molten pool formation in the reactor vessel. It may be considered as an investigation of corium configurations that were not considered in previous programs such as the OECD RASPLAV and MASCA programs. All experiments within the program are carried out in the Rasplav tests facility of NITI, using the technology of induction melting in a cold crucible. This program is co-funded by IRSN, EDF, AREVA and CEA in the frame of the CPSIN agreement. This paper presents the main conclusions drawn from CORDEB tests, both for phenomenological knowledge and for transposition to reactor situations. Quantitative elements are provided for a clearer picture of kinetic effects. CORDEB tests have shown the importance of chemical interactions between molten steel and solid sub-oxidized (U-Zr- O) corium which lead to penetration of steel inside the solid oxide (crust or debris) and to the partial dissolution of the solid. As a consequence, the crust surrounding a molten pool cannot be considered as a barrier to mass transfer. CORDEB tests also made it possible to assess the oxidation kinetics of a molten pool: in the presence of surface crust, this is a rather slow process and it does not have a significant effect on the deterioration of accident scenario. The gaseous atmosphere above the pool plays an important role in several processes, in particular in the evaporation of molten steel.