Causal research, for the most part, treats fear as a characteristic of individuals and examines how it is affected by other individual characteristics, such as age, sex, class, and race. Our research treats fear as a characteristic of social units (cities) and examines how it is affected by the structural and cultural characteristics of those units, such as crime rates and the newspaper coverage of crime. The sample consists of the 26 cities used in the National Crime Survey (NCS). Data on the fear of crime are obtained from the NCS; data on the structural characteristics are obtained from a variety of sources, including the Uniform Crime Reports (UCR), NCS, and the U.S. census; and data on newspaper coverage are obtained from a content analysis of the newspapers of the 26 cities. The results show that the effect of newspaper coverage is complex, with some forms of coverage increasing fear and other forms of coverage decreasing fear, and that the effect of official crime rates is mediated through the newspaper coverage of crime. 4 tables, 1 appendix, and 31 references. (Author abstract)