Two experiments examined the effect of group discussion on subsequent evaluations of job applicants. The hypothesis was tested that discussion would polarize the evaluations of applicants such that the average post-discussion evaluation would be more extreme in the same direction as the average of the prediscussion evaluations. The relationship between polarization and quality of the evaluations was also examined. Measures of evaluation quality included interrater reliability and accuracy of recall of applicant and job characteristics. Subject groups were employed under two Discussion conditions: discussion of applicants or discussion of an irrelevant topic. Experiment 1 found a marginal polarization effect and an increase in interrater reliability following discussion of applicants. Experiment 2 also found a marginal polarization effect but no differences between conditions on interrater reliability or recall accuracy. Combined results suggest discussion may serve to polarize evaluations of job applicants, although each individual experiment did not statistically confirm the hypothesis.