A survey concerning international harmonization of accident reporting was distributed to 80 experts in accident reporting and analysis. Completed surveys were received from 50 persons in 13 countries; 74% of the respondents had more than 10 years of experience in the field of traffic safety. The main findings of this survey are: (1) 86% of the respondents think that an international computer file of disaggregated fatal-accident data would contribute to understanding of traffic safety, and 84% would use such a file. (2) An international non-fatal-accident file was considered to be of value in research on human factors and accident causation (60%), and in determining black spots in the road network (57%). (3) Police was the most frequently mentioned source of data for both the fatal and non-fatal international data files. Nevertheless, fewer than one-quarter of respondents considered police as the suitable exclusive source of either data. (4) The majority view was that the data for both types of files should come from more than one agency. (5) In the case of the fatal-accident file, 78% of the respondents considered it important that the data be cross-checked with the public health records. (6) The 10 most useful variables for a fatal-accident file were traffic unit type (e.g. car), accident type (e.g. angle), road class, driver age, date/time of day, age of person killed, number of killed persons, number of injured persons, drinking or drug use, and restraint usage of person killed.(ABSTRACT TRUNCATED AT 250 WORDS)