Methods and reasons for evaluating teaching are discussed, and an experimental study of the effectiveness of students' ratings of teachers is described. The two main reasons for evaluating teaching as given in this paper are (1) to help make decisions about whom to promote, and (2) to improve instruction. In the experimental study, five diverse colleges participated. A total of some 470 faculty members were randomly assigned within each institution to one of three groups--feedback within a week (treatment group); no feedback, with summary of results given at end of the semester (control group); and posttest, which used rating form only at the end of the semester to determine whether simply using the form caused teachers to change, even without feedback. A 23-item form eliciting instructional procedures or behavior that an instructor could presumably change was used in the study. Results showed that instructors who received student feedback did not noticeably modify their teaching practices. A second aspect of the study was to determine to what extent instructors describe or rate their teaching differently from the students' ratings. Items from the student form were reworded slightly for instructor responses. It was found that there was a significant difference between instructor and student responses to most items, with instructors rating their teaching in more positive terms. The use of student pre- and post-test scores as a means of evaluating the effectiveness of teaching are seen as beneficial to the teacher, but their use as the sole criterion for determining teaching effectiveness is not advocated. Suggestions are made as to other evaluation techniques.