This article suggests a method to make estimates of the risk of detection based on information from the Swedish crime statistics. The risk of detection is expressed as a function of the dark figure. Empirical estimates of the risk are given for drunken driving, residential burglary, and assault between strangers. These estimates are followed up with some calculations concerning the impact of crime activity level on the risk of getting caught. One major conclusion of the findings is that the risk of detection varies very moderately with the dark figure. This means that if there is some knowledge of the dark figure, it is often possible to make fairly good estimates of the risk of detection. 3 figures, 3 tables, and 17 references. See NCJ 124605-124612 for individual papers. (Author abstract)