The dependent study variable was the city homicide rate, while the measure of welfare assistance was the average AFDC payment amount per recipient person for each State in which a sample city was located. The AFDC payment per recipient person, adjusted in accordance with cost of living, had an independent, direct negative impact on homicide rates and a separate, indirect negative relationship to homicide rates through its association with household status. Results support both strain and control theories of homicide and, in particular, Sampson and Wilson's social disorganization-strain perspective. Further, results imply that lowering welfare support levels for the poor may elevate homicide rates and that increased differences among States in AFDC assistance levels may increase interstate variations in homicide rates. 39 references, 1 table, and 1 figure