This study investigated whether readers of newspaper crime reports make assumptions about the racial identity of criminals from news accounts that make no explicit race references. Simulated crime stories given to respondents were experimentally varied on the factors of (1) urban character of the crime location (urban versus suburban), (2) ethnic character of the crime location (white versus nonwhite), and (3) type of crime (violent versus nonviolent). Results indicated that: (1) urban character and ethnic character of crime location interacted to produce relatively high tendencies to attribute crimes to blacks in all given locations except the suburban/white area; (2) violent crimes were generally attributed to blacks, while nonviolent crimes were attributed to whites; (3) more whites than blacks tended to attribute crimes to blacks; and (4) neither dependence on mass media (newspapers or television) nor knowledge of the race of criminals in previous victimization experiences affected the respondents' tendency to attribute crimes to a certain race. (Author/MJL)