The final objective of all severe accident evaluations is to obtain th
e magnitude of the potential Fission Product (FP) release in the envir
onment. Consequently, severe accident experts have focused, very often
mainly, and sometimes exclusively, on FP physical and chemical behavi
our. In many evaluations, the thermal hydraulic transient was calculat
ed by very simplified models. But the situation is evolving. By carryi
ng out more frequent physical analyses to improve plant evaluation or
to prepare experiments (such as Phebus FP), a growing number of therma
lhydraulics problems have been raised and recognized as being signific
ant for FP prediction. The paper therefore reviews the main thermalhyd
raulic phenomena which have to be taken into account and especially ex
amines their influence on severe accident initiation and scenario, cor
e degradation, vessel failure and more generally RCS failure, FP relea
se, transport, deposition, revaporization. The needs for model improve
ment and for specific assessment are listed. They cover not only the d
escription of special thermalhydraulic phenomena, but also the handlin
g of complex coupling with other physical modules. Prospective researc
h directions are proposed to answer these needs.