[en] Many probabilistic risk assessments (PRAs) provide little documented information on either nondominant accident sequences or component failure combinations (i.e., cut sets). Therefore, it is difficult to use a published ORA to evaluate alternatives for improving a plant's safety. Criteria are developed for defining an appropriate number of dominant accident sequences and associated cut sets which should be published in a PRA. Applying the criteria, it is illustrated for one power plant that only 1116 component failure cut set out of 199,600 total cut sets need to be documented to assure that undocumented accident sequence probabilities add up to no more than 2.5 x 10^-₅ per reactor year