The study assumed that more positive images of the police would be reported by older, white, married females who were highly educated and held white collar occupations; that more negative images would be reported by respondents and their personal aquaintances who had trouble with the police; and that more negative images would be reported by respondents who lived in high-crime census tracks than those residing in low-crime census tracks. The study sample consisted of 621 randomly selected respondents from four predominantly black census tracks, two with the highest crime rate (180 and 200 index crimes per 1,000 population) and two with the lowest crime rate (40 and 60 index crimes per 1,000 population). Data were collected via personal interviews. Study results showed that a larger proportion of black Atlantans rated the police as more honest but less effective in delivering community services than did blacks at the national level. In addition, residence was the single best predictor of police image, followed by marital status. The author recommends that policymakers interested in improving the police image focus on young, single residents in high-crime census tracks. A more positive police image within this group may improve their cooperation with the police and result in crime reduction and prevention. 37 references, 4 tables, and 1 figure