The effect of instructing subjects to respond slowly or quickly to items that measured perceptions of defendants accused of committing a race-stereotypic or a race-nonstereotypic crime was examined. Consistent with findings from previous related research, subjects judged defendants accused of race-stereotypic crimes more negatively than defendants accused of race-nonstereotypic crimes. As predicted, the interaction between defendant race and type of crime was affected by the manipulation of response time (RT) instructions. Subjects who were instructed to respond quickly provided responses that were consistent with the previously documented interaction. in addition, the perceived severity of the crime and attributions regarding the defendant's behavior mediated the relation between perceived typicality of the offense and punishment recommendations. The data revealed that these mediational effects were attributable to those subjects who received the quick RT instructions. The findings are discussed in terms of the distinction between the activation and application of stereotypes. The implications of these findings for stereotype research that uses response latency as an index of category accessibility are also addressed.