An attempt is made in the first part of this paper to review the history of reactor safety and draw conclusions about trends that could be avoided and directions that could lead to robust reactor designs that would not be susceptible to severe accidents. In the second part, progress in reactor thermal hydraulics is observed by reviewing the list of conference sessions. Finally, a report is made on some recent work on two computational problems: the prediction of departure from nucleate boiling and the potential spatial coupling of computational multifluid dynamics methods to achieve multiscale, high-resolution simulations.