The analysis focused on the rate of police killings in 1980. Data came from supplementary homicide reports that police departments filed with the FBI. Since police killings are unusual events, police killing rates were analyzed over 7 years, starting in 1980 and ending at 1986. The hypotheses tested involved political-threat theories and reactive hypotheses. Under political-threat theories, police may use deadly force because they protect the interests of the privileged by keeping the redistributive violence of subordinate racial or economic groups in check. If the "economic" version is correct, the most economically unequal cities will have the most police killings. If the "racial" version of this political-threat approach is correct, the following four relationships should hold: Police killing rates will be higher in cities with more blacks or in cities where growth in the percentage of black residents has been pronounced; greater use of lethal force can be expected where there are larger differences in the economic resources of blacks and whites; and the police should not use lethal force as often in cities with a black mayor. Under reactive hypotheses, the police may kill because they must protect all groups from violence. The independent variables measured reflected these categories of hypotheses. Tobit analyses of the 170 cities shows that racial inequality explained police killings. Interpersonal violence measured by the murder rate also accounted for this use of lethal force. Separate analyses of police killings of blacks show that cities with more blacks and a recent growth in the black population had higher police killing rates of blacks, but the presence of a black mayor reduced these killings. These findings support latent and direct political-threat explanations for the use of lethal force to preserve order. 3 tables and 47 references