Logistic regression models were estimated using the 1994 General Social Survey to examine effects of race, gender, and geographical region on support for various criminal justice policies. Dependent variables included support for capital punishment, perceived harshness of courts, and support for police use of force under five conditions of escalating severity. The total sample size was 2,992 (83 percent Caucasian, 13 percent African American, and 4 percent "other", primarily Hispanics and Asians), 43.1 percent male, 56.9 percent female, and approximately 36.6 percent from the region coded as South. African-Americans were less likely to support capital punishment and police use of force than their counterparts, results that were enhanced when re-estimated using only the southern sample. Thus, regional prejudices affected minorities' perceptions of bias in criminal justice policy and practices. Although African Americans were less likely to endorse police use of force in general, the study found situationally specific instances in which African Americans did endorse police use of force. Tables, references