Devine (1989) has proposed a model which suggests that when individuals are required to evaluate stereotyped groups both low-prejudice and high-prejudice persons will automatically activate the pertinent stereotype content. Controlled inhibition of the content of the stereotype characterizes low-prejudice responses and distinguishes low-from high-prejudice persons. The experimental evidence Devine (1989) reports to support this model is criticized on several grounds, and a methodology that avoids these shortcomings is described. Two groups of subjects were tested, one with and one without a stereotype of the target group. The results from the group knowledgeable of the stereotype were inconsistent with Devine's (1989) model. Differences in the automatic activation of stereotype content, rather than differences in the tendency to employ strategies to inhibit stereotype content, were related to levels of prejudice towards the target group. The results from a sample of subjects not possessing a stereotype of the target group confirmed the integral role of automatic processes in prejudice. Prejudice for these subjects was related to the tendency to automatically activate negative rather than positive concepts, irrespective of their relatedness to the stereotype of the target group. The implications of these findings for Devine's (1989) model and the more general nature of the relationship between stereotypes and prejudice are discussed, and the advantages of the current methodology detailed.