In a randomized controlled trial three methods of feedback of increasing intensity, directed at 183 doctors taking cervical smears, were compared with respect to their effects on the sampling quality of smears. Overall, feedback was found to have no influence on quality criteria in the crude data analysis. However, a significantly larger decrease in the percentage of smears lacking endocervical cells was found in the groups receiving monthly overviews of their results with peer comparison, when compared with the groups not receiving this type of feedback (odds ratio 0.75). Moreover, feedback appeared to have a clear effect on the presence of endocervical cells among doctors submitting a substantial number of smears in the intervention period, as opposed to those who submitted fewer smears. A positive correlation was also observed between the increase in the group mean of the proportions of smears containing pathological cells and the intensity of the feedback. However, this increase did not reach statistical significance. This study suggests that monthly feedback with peer comparison may have a positive relationship with some aspects of quality improvement in cervical screening.