[en] The success of probabilistic risk assessment (PRA) has led to numerous improvements to the safety of commercial nuclear power plants. Those very successes, however, have led to a situation where the events that PRAs have traditionally modeled are extremely rare. As a result, current PRAs have the potential to be misleading because events previously considered to be unimportant and, thus, not included in the scope of the analyses may now be dominant. This paper provides an assessment of the current status of accident sequence analysis methods and identifies particular limitations that should be addressed in future studies