Survey data was gathered during 1974-75 from a study sample of 250,000 persons. In structured interviews, respondents were asked to estimate the number of offenses reported to the police for each 17 kinds of offenses. To promote maximum recall among respondents, researchers always questioned respondents about the preceding year. Each respondent was presented with only four of the offenses, and offenses were presented using a reticulated design so that each offense was evaluated by an equal number of respondents. Offical data were obtained from tabulation of the computerized offense records made available by the metropolitan police department. The relation between the perceived incidence and the official incidence of the offenses is described by a power function, the parameters of which are highly replicable across the surveys. Although there is systematic distortion, the overall degree of accuracy was found to be remarkable. The study's findings of a remarkable correspondence between official information and public perceptions run counter to the literature on media coverage of crime. The study suggests that the accuracy of public beliefs is a result of the fairly accurate depiction of the official incidence of serious offenses by the media. Because of its importance, the information enjoys a high degree of diffusion as well as retention among the public. Two tables, a figure, and footnotes are included; 30 references are provided. (Author abstract modified)