The basic problems involved in the safety evaluation of power reactors were reviewed. The present days evaluating practice in Japan is considered satisfactory for the basic requirements in solving the sociological sides of reactor hazards. However, when we will proceed from subjective judgements on case-by-case basis, and give a direction to subjectivity of judgements, and further, when we intend to get a quantitative expression as site evaluation criteria, the most important problem involved will be the way how to express the empirical probability concept.
In this paper, the safety evaluation procedures taken in over-seas countries were reviewed from this standpoint, and taking the peculiar conditions in Japan into consideration, it is attempted to find a direction in which we should proceed. The ground rules or the path to be followed in safety evaluation is proposed, to introduce a consideration of probability of more serious accidents into the present practice in Japan. But, research and development efforts in various areas are to be made before we could go further from the groud rules to the quantitative expression, i.e. the site evaluation criteria.
