Changes in teacher behavior in response to different types of feedback were examined. Subjects were faculty members from a two-year college. Five types of feedback were selected: (1) videotapes made during one class period; (2) the Bellack system, which provides the teacher with an observer-made graph of frequencies of teacher and student interaction; (3) every question asked by the teacher during a given period written down verbatim; (4) written reactions by students on specific teaching activities during instruction; and (5) questionnaires gathering student perceptions of the class, teacher, and their own progress. Teachers' perceptions of a feedback's novelty, credibility, and relevance were also sought. Comparisons among the five types of feedback showed that the types most related to student learning increased teacher change to a greater extent than did types less related to student learning. Teachers did not perceive the five types of feedback differently on the dimensions of novelty, credibility, and relevance. Student questionnaires were seen most likely to produce change, followed by students' written reactions on class activities. Videotaping proved to be less associated with teacher change than did other types of feedback. (JD)