This follow-up study of the selection of residents in Topeka produced several conclusions. The appraisal of three interviewers is about as accurate as appraisal derived from psychological tests. Appraisal by a single interviewer is unsatisfactory and but little more useful than selection by chance. Two interviewers were better than one and three than two; apparently they tend to cancel out and compensate for each other's errors. The accuracy of prediction has no clear relationship to the age or experience of the interviewer. There is a tendency to underestimate applicants who show much anxiety or aggressiveness and to overestimate those who show evidence of cultural attainments. No single procedure was found that can supplant or consistently improve the interviewer's clinical