This paper discusses how the long-term frequency of major transportation accidents can be estimated for a small country such as Norway, for which the observed frequency may not be a good estimate for long-term frequency. A major transportation accident is defined as an accident in which there are at least five fatalities. Estimates of the long-term expected frequency of major transportation accidents in Norway are developed by combining information from three sources. They consist of historical records: for Norway for the period 1970 to 2001 (32 years), for Great Britain for 1967 to 2001 (35 years), and for Europe for 1991 to 2003 (12.5 years). The use of formal techniques for decision analysis to help in setting policy priorities in prevention of major transportation accidents is discussed. Also, trade-offs between three safety-related policy objectives are discussed: preventing major accidents, reducing differences in accident risk, and seeking a maximum reduction of the total number of accidents. A trade-off between these policy objectives cannot be made by reliance on a single monetary valuation of safety, but it can in principle be made by treatment of the objectives as attributes of a multi-attributive goal structure that can be maximized by application of multi-attribute utility analysis. Utility values obtained in such an analysis in turn imply a set of monetary valuations reflecting the relative importance of the different policy objectives.