A study was conducted to evaluate certain formats for providing feedback of student opinion during teacher training. A class of 69 interns were randomly assigned to five treatment groups and one of their classes was chosen to be used in the experiment. In Febraury each class in three of the groups was given the Bryan Student Opinion Questionnaire (SOQ) and a parallel self-appraisal form only, and a fifth group was not tested until the end of the experiment. The three groups of teachers whose pupils were given the SOQ were given three feedback treatments; (1) no feedback, (2) written feedback only, and (3) written feedback plus a conference with a supervisor. In early May all five groups were given the SOQ and the self-appraisal form. The data were analyzed in three stages: (1) to compute basic statistics, correlations, and factor analyses, (2) to detect differences between treatment groups, and (3) to detect possible sources of invalidity and to assess the accuracy of self-appraisal. There were significant differences between the treatment groups. Feedback, as compared with no feedback, was effective in changing subsequent pupil ratings, but in a direction opposite that which was expected. Those teachers receiving feedback were rated lower in May than in February. The conference mitigated this effect, but did not eliminate it. (Findings, instruments, and references are included.) (JS)