173 undergraduate students were randomly assigned to 4 groups. Group 1 (n = 43) received a description of general objective criteria for the job of police officer. Group 2 (n = 43) received additional information describing variables relevant to the valid prediction of job performance and were instructed to pay attention to those variables. Group 3 (n = 46) received additional information describing those variables not valid for job selection and were instructed to attend to those. Group 4 (n -41) received the descriptions of and instructions to focus on both relevant and irrelevant variables. All subjects then rated 4 fictitious job applicants on the basis of application materials that were manipulated in reference to both their relevant/valid qualifications and their irrelevant/stereotypical fit to the job. Analysis indicated that subjects attending to irrelevant information rated applicants more objectively. Assuming these results generalize to cognitive processes that occur among actual personnel managers, findings suggest that training raters to attend to rather than ignore irrelevant information may help reduce bias in personnel selection.