In previous research, targets' sensitivity to prejudice cues has been assessed on the basis of two types of information. Prototypical information renders the situation representative of discrimination encountered by the ingroup. Diagnostic information is a direct indication that prejudice possibly is operating in a given situation. We hypothesize that, when available and processed at the onset of an evaluation situation, prototypical information shapes targets' understanding of subsequent diagnostic information. In three experiments, participants were informed that they were to be evaluated by relevant outgroup members either before or after having performed a task. Diagnostic information was always provided at the same moment, i.e. after the task was completed, and was either uncertain (prejudice may bias the evaluation) or certain (prejudice certainly biases the evaluation). In the before condition, attributions to prejudice were as elevated whatever participants were told that prejudice might, or certainly did, bias the evaluation. Furthermore, in the case of uncertain diagnostic information, those who were readily informed of the evaluator(s)' identity attributed their failure to prejudice to a greater extent than those who received this information later. Copyright © 2004 John Wiley & Sons, Ltd.