A probabilistic approach to engineering advances. Since risk adhering to a technical structure can be determined quantitatively easier and more accurately than before, criteria for decision making are becoming more important. If the structure is in the public domain, and benefits are not felt, the voluntariness of being subjected to the risk becomes low. Decisions become, in such a case, a political issue on the basis of considerations of ethics, law, and social justice. For a specific case, the comparison of the economic benefits with the total costs, including risk reduction in a worst case scenario, play a major role as we have seen over and over again in, for example, planning nuclear power generation. A criterion for individual risk of 10(-5) per year for the purpose of "external safety" is generally accepted. For social risk assessment, however, a group risk criterion is essential and much more difficult to define. An analysis is made. Examples of inconsistency are given. Further dialogue is encouraged. (C) 2003 Wiley Periodicals, Inc.