Using Swedish rape statistics as a focus, the article claims that cross-national comparisons of crime levels are extremely hazardous when based on official crime statistics, since the construction rules vary widely. International comparisons of crime levels should as a rule be confined to findings of international victim surveys. The example of rape statistics in Sweden--about three times higher when compared to other countries in the European Sourcebook--is used to explain what factors can influence statistics. The article shows that changes in statistical routines, the legal definition of rape and changes over time all influence the statistics in a substantive way. The article also indicates the great extent to which crime statistics are a construct whose appearance is very sensitive to the rules applied in the process of construction. Using statistics appropriately requires a thorough knowledge of the principles guiding that process. Tables, notes, figures, references