This article deals with the assessment of severe accident event sequence and mitigation strategy elaboration performed in the frame of ASTRID SFR Gen IV reactor project (Le Coz et al., 2013), including the Japanese/French collaboration involving JAEA, MFBR, Framatome and CEA. In the first part of the paper, the mitigation strategy and the approach defined for the design of mitigation features, which are transverse tubes called DCS-M-TT are presented. By considering 21 DCS-M-TTs into the core as defined at the end of the conceptual design, which enable to extract the core materials and to significantly reduce the core reactivity within the order of 10 s in some typical accidental conditions, the accidental power excursions in ASTRID should not lead to a significant mechanical energy release, thus ensuring the integrity of the main vessel. Nevertheless, some loading assumptions independent from event evolutions have been defined in order to design the main vessel and the core catcher in order to cover extreme accident consequences and to get rid of event sequence dependency when designing the reactor. Then, SIMMER-IV (3D) the main calculation results of ULOF simulations whose goal was to verify the DCS-M-TT efficiency and to support a PIRT are highlighted in the paper. They confirmed that fuel discharge through discharge tubes has a large mitigating impact on power excursions. The last part of this paper deals with an approach dedicated to the description of severe accident event sequences through generic event trees (GETs), which logically summarize the representative courses of accident progression showing relationship between critical causal events and major outcomes of the accidents. A set of trees has been elaborated for the description of each event sequence able to cause severe accidents (ULOF: unprotected loss of flow, UTOP: unprotected transient overpower and USAF: unprotected sub-assembly fault). The event-tree bifurcation points towards one branch or another of the possible event evolution could be examined alternatively with a help of phenomenological event chart (PECs). Afterwards, in association with the GET analysis, a PIRT (Phenomenon identification ranking table) on ULOF has been developed conjointly by Japanese and French sides in order to prioritize R&D needs. About 100 phenomena have been retained and classified depending on the various time periods of the accident event sequences; the primary phase, the short-term relocation phase, the expansion phase and the long-term relocation and cooling phase.