The objective of this study was to determine whether formal, written feedback to teachers would improve the quality of teaching. Twenty-seven units that were teaching clinical skills in teaching hospitals (TH), district general hospitals (DGH) and GP surgeries were used as the setting. One hundred and eighty-eight students in Audit Phase 1 (1997) and 175 students in Audit Phase 2 (1998) completed a questionnaire regarding time utilization during teaching sessions, tutor characteristics and course attributes. Feedback of results from the Audit Phase 1 questionnaire to each individual unit was combined with anonymized results of the other units. TH units (which had scored poorly in the first questionnaire) significantly improved the quality of their teaching. This was seen in increases in (1) the degree of prior preparation, (2) the number of sessions that adhered to the guidelines provided in the study guide, (3) time spent with patients and (4) the interest shown in the students. By contrast, the DGH units and GP units (which had scored well in the first questionnaire), scored poorly in the second questionnaire for prior preparation (GP units) and educational value (DGH units). It is concluded that feedback can improve the quality of student teaching. However, when presented as a contrast with other poorly performing units, feedback may induce complacency.